Election Oversight, June 2016

Direct link to this page: http://www.copswiki.org/Common/ElectionOversightJune2016
Share Button

Latest News

This topic is now being used as a basis for more general instructions, here: Election Oversight Procedures

The primary election has many new developments which cause concern, and could flip the election in favor of Bernie Sanders once and if all the ballots are counted.

Here is the situation: In this election we had many new voters and many nonpartisan voters who voted on a nonpartisan ballot. In California, parties can choose to allow voters to vote in their primary if they are nonpartisan voters. Republicans do not allow this but Democrats do and so most other parties. So nonpartisan voters with the NPP designation can choose the DEM ballot or another party ballot (but not Republican). In this election, the vast majority of such NPP voters will be Bernie Sanders voters.

Some NPP voters had trouble getting the DEM ballot so they could vote in the presidential race, and were given a pure nonpartisan ballot (with no presidential voting options). If they did not know enough to ask for the right ballot, then they may be mistakenly vote without even being able to vote for president. So that is one issue we should be aware of and try to get an idea of how rampant this was.

Normally, NPP ballots are supposed to be placed in the same box as other (DEM, REP, etc.) ballots. However, in this election they started to group these as "provisionals," which is normally for say voters in the wrong precinct, voters who were vote-by-mail (VBM) but did not have a ballot to surrender, etc. The number of provisional is far higher than any other election as a result. A precinct may have a half dozen provisionals. In this election, there may be hundreds per precinct.

Another name for provisional ballots could be "challengeable ballots" because these become challengeable by anyone. It's really an outrage that an otherwise legit voter who simply chooses a party ballot -- which is normal and expected -- suddenly becomes a provisional (challengeable) voter.

Each precinct must log in a provisional ballot on a log sheet and explain the reason why. Reviewing these log sheets will be very important in ensuring that no NPP/DEM ballots are lost and are reflected in the final vote.

You have the right to review the processing of the provisional ballots. In most elections, the typical approach to processing provisionals is to try to get as many allowed for your party while not allowing those of your opponent's party. This review is done before opening the envelope and you cannot look at the ballot. For example, if a VBM voter comes in and says he could not find his VBM ballot but did not mail it in and wants a new ballot, then in the review process, they would verify that no VBM ballots were received prior to counting that provisional. VBM ballots would need to be processed before a final conclusion on this ballot could be determined.

Typically, the candidates will split the vote R vs. D. So the D candidate will try to exclude R voters and include D voters, and vice versa. This should not apply to nonpartisan voters who simply choose a partisan ballot. There should be NO further review or opportunity for anyone to challenge NPP ballots unless there is some other legitimate reason these should be provisional and thus challengeable ballots. If there are H activists there trying to challenge NPP/DEM ballots in anyway, we must be very assertive that this should not happen!

So the provisionals are processed after all other ballots are processed due to the VBM double-voting issue described above (or they may simply set aside those that were VBM voters who did not have a ballot to surrender). If the NPP voters who requested a DEM ballot are grouped with provisionals, this will delay processing of those to the very end, and can mean that an election, like the current one may flip, esp. when it seemed to many poll workers that 80% of the voters were nonpartisans voting this way.

--> We must make sure those voters are not disenfranchised by calling the election early. In this election Bernie Sanders should not concede or drop out until the official results are submitted 29 days after the election.

Background

You may want to review the work we did in 2010 to fully review the procedures used at the San Diego ROV. See the summary here: San Diego County Canvass Procedure There is a link on that page to the full report.

Instructions for Oversight Volunteers

To be the most productive, we recommend the following:

Daily vote count capture

Each day, you should capture the website results from the registrar's website. In San Diego, they use an XML format. You should save the raw XML file for later review as it contains information that is not revealed in the rendered version. For ease of use, I also capture the top of the page as a JPG (using "Snipping Tool") which shows the total ballots still uncounted and the vote counts in the presidential race. Save these for later use.

SAMPLES

NOTE It is most important to capture the totals PRIOR to any processing of provisionals because we need to get an idea of how they are affecting the race.

Provisional Ballot Processing Oversight

Provisional ballot processing is slightly different than usual in this election. As described above, our main agenda will be to make sure NPP voters ballots are immediately reclassified as legit ballots and no further challenges should be allowed (unless there is some reason noted for allowing a challenge, like wrong precinct, etc.) The main thing we can do here is to total up the number of NPP/DEM voters, who are likely Bernie Sanders voters who were improperly sidelined into the provisional stack.
  1. Make sure NPP ballots are not challenged unless there is some obvious reason noted.
  2. COUNT all NPP and NPP/DEM ballots. If the registrar is already creating a count in their process, then it is sufficient to make sure they are and to make sure it is accurate and to check the process now and then. If they do, then this may not require 100% oversight by volunteers, and occasional checking may be enough. If they do not, it is optimal if we can make our own count for a check on the results (as they can be modified later due to the possibility of compromised workers or hackers).
  3. We should also count how many people were given nonpartisan ballots without a presidential race. It is really not that likely that NPP voters would really not want to vote in the presidential race, so these are likely one more sleight of hand in suppressing the vote. This will help us understand the first issue identified in the longer description above.

Precinct Roster Review Procedure

Capturing information from the Precinct logs may be much more important than just making sure the NPP ballots make it through the screen unchallenged.

Background

At the precinct, workers are supposed to log in every single voter who is processed. One of the logs is for provisional ballots. We have attempted to get these scanned and turned into images so we could review them later on the website. But they resisted because there were signatures on the pages. So the best we could do was to sit down with binders and transcribe the information into a spreadsheet that we could review later. In our review in 2008, we did 85 precincts, a 5% sample. If you chose to do a sample, you should chose the first 1% as the same precincts involved in the 1% manual tally, then include another 4%. If we have enough volunteers, do another 5%. Mainly, in our earlier review, we counted how many voters were in each of the many log pages for each precinct.

1. Make an appointment with ROV

It will be necessary to make an appointment with the Registrar of Voters to look at these materials.

2. Prepare data logging sheets.

The basic fields we used in our prior review are as follows. (Assuming no changes to the forms).
Field Enter data from this field
Blank Ballots Ballot Statement (1), hand count before election
Unused Ballots Ballot Statement (2), hand counted after election
Voted Ballots Ballot Statement (3), from ballot box
Provisional Ballots Ballot Statement (4), number of ballots in peach envelopes
Spoiled Ballots Ballot Statement (5), number of spoiled ballots
Blue Page Signatures Ballot Statement (7A), number of signatures on Blue pages
White Page Signatures Ballot Statement (7B), number of signatures on White Pages
Pink Page Signatures Ballot Statement (7C), number of signatures on Pink Pages
Peach Provisional Signatures Ballot Statement (7D), number of signatures on Peach pages
Total Signatures Ballot Statement (7E), total signatures as reported
Touch Screen Ballots Ballot Statement (9), number of ballots completed on touch screen DRE machines
Total Voted Ballots Ballot Statement (8A), Add lines 3, 4, 9
Ballot Statement Notes Ballot Statement (8B), do values match and why
Ballot Count Adjustment Our adjustment to get to our best estimate of ballots in the box
Adjustment Notes Why an adjustment is valid.

In addition, we want to get additional information by reviewing the signatures and reasons on the provisional logs sheets.
  1. number of NPP/DEM provisionals
  2. other provisionals, esp. those that do not have any presidential option.
  3. Unfortunately, the precincts in San Diego DO NOT keep track of the number of VBM ballots dropped off at the polling place. But if they have fixed this, then keep track of this too.

You can see the captured data from the work we did in our 5% sample in 2010: http://www.copswiki.org/Cops/AllPrecincts

Here is a TUTORIAL VIDEO:

It will likely be best to use two volunteers for each precinct, one to count and one to take the information, either on paper or just enter into a spreadsheet. I am estimating perhaps 10 minutes per precinct. If you did it for all 1522 precincts, it would then take 1522*10min = 500 person hours. If we had ten people working diligently, this would still take more than a week (50 hours).

The best way to work is to start with a 5% sample, then go to 10%, etc. Doing a 10% sample will still take 50 person-hours. Ten people could do it in 5 hours. For 1522 precincts, that is 152 precincts and would be a really good set of data.

In San Diego, June, 2016, the following precincts (sequence numbers) were chosen by random draw in the 1% manual tally selection process.

1% Manual Tally PrecinctsSorted ascending Additional for 5% Additional for 10%
32, 363, 368, 418, 597, 637, 670, 686, 857, 877, 991, 1229, 1332, 1418, 1431, 1454 TBD TBD

We recommend that you do not inform the ROV until the last minute what your additional precinct numbers are so they can't fix up those rosters in any way.

NPP/NPP and NPP/DEM Investigation

In the 2016 primary, we discovered a vast number of NPP voters who were not given the opportunity to cast a ballot with the presidential race on it, despite the fact they have a right to do so to some parties.

PROCEDURE 1 -- Phone Number & Address Harvesting

This step can be separated into a separate task processed by a separate team. There is no phone calling in this task.

SEQNUM NPP/DEM NPP/NPP Name Phone Address CityZip Q01 Q02 Q03 Q04 Q05 Q06 Q11 Q12 Q13 Q14
13   1 Matthew Adams   181xx SENCILLO DR SAN DIEGO 92128                    
13   1 Lloyd Billotte xxx-xxxx 122xx GREENS EAST RD SAN DIEGO 92128                    

PROCEDURE 2 -- Phone Calls

TELEPHONE SCRIPT

Hi, this is (your name) from Citizens Oversight, a nonpartisan nonprofit organization, and we are conducting a survey of election anomalies in the June Primary Election. Am I speaking with (name on the roster)? Do you have a minute to answer a couple of questions? Any information we gather is confidential.

--NO-- Is there a better time to call? (record in Call Status CB:date)

NP/NP Voter

NP/DEM Voter

Official canvass file

It will be necessary to obtain the nearly official canvass file, broken down by precinct. DO NOT wait to the very end to get these.

The files of this type are text CSV data files and tend to be very large, and only provide a single data item per file with other columns being indexing descriptors. There are many columns that provide redundant information. If you boil it down to the minimal descriptors, you get something like:

Precinct Race Ballot Option Type of Ballot votes
(the number, place, etc.) (ex: President) (candidate or yes or no, etc.) Polls vs. Mail (number)

They also provide the number of ballots where there are under or over votes for the race as separate line items. This file tends to be quite large, several 100 MB.

We have a (Perl) script that will condense this file down so there is one line per precinct and many columns for each ballot option with totals in each column.

1% manual tally oversight

Observe the 1% manual tally.

Chain of Custody Oversight

We got news that poll workers were taking boxes of ballots home with them.

Data processing

A key question to answer is whether the NPP/DEM ballots are not getting lost somehow between the precinct log sheet and the final totals. PLEASE NOTE: the official data file does not contain any information regarding the source of the ballot in this regard. The best way to check this will be to note the total counts prior to provisional processing and see if they go up according to the NPP/DEM ballots processed.

Snapshot Protocol

Citizens Oversight has conducted an oversight procedure for a few years called the Snapshot Protocol. In this procedure, we ask for partial (but as complete as possible) results prior to the 1% manual tally random selection in the form of a canvass data file. This protocol can catch most central-tabulator manipulations. But unfortunately, it generally excludes the provisional ballots that are so critical in this election. This snapshot of data must be provided to us (or a third party) prior to the 1% manual selection. We asked ALL counties in California to upload their snapshot file to our server.

We need some volunteers to go through the status of the snapshot data file in each county. And then, to close the loop, we should compare the 1% manual tally counts with the snapshot file, to confirm that no compromised election worker or hacker modified the 1% manual tally computer results to match the paper ballots.

Some issues to look for

MISSING DATA
One of the issues in San Diego is the fact that the snapshot data file does not include all the ballots, but only polls ballots. VBM ballots and provisionals are grouped into "batches" that are not directly reflected in the canvass data file. These batches are mixed precincts and ballot types and are directly read into the central tabulator with no paper trail. The registrar does a manual tally on these batches and they compare with a computer file of some sort. We have asked for the data file which is used to verify the 1% count of these mixed batches.

INSUFFICIENT DATA
State law says that election districts in California must perform a 1% manual tally. However, in San Diego, they interpret this as 1% of the ballots processed as of the end of election night and exclude some 285,000 ballots still needing to be counted. The number of batches selected for the 1% manual tally is really only about 0.5% of the total (one half of one percent) because they are ignoring the uncounted VBM and provisional ballots. We have requested that they comply with state law.

IN ADDITION, we would prefer it if they include ALL the ballots in the count prior to the 1% manual tally and not rush into it. The way they do it in this county makes me believe their agenda is to allow election manipulation.

CHAIN OF CUSTODY VIOLATIONS
In voting system that use a durable paper ballot, the chain of custody is extremely important. We need to not only know if ballots had any chance to be changed in the ballot boxes, but also the total counts of printed ballots. There should be no unaccounted-for printed ballots that were not either voted, spoiled, or returned. (otherwise, they can remove ballots and replace with ones voted the way they want. The number of seals that are broken or missing. There should be documentation that goes along with the seals that includes the number of each seal used. Any unused seals must be accounted for (otherwise, they can just remove the seal, and seal it back up with another one.)

CENTRAL TABULATOR AUDIT LOG
The central tabulator log should be reviewed and inconsistencies explained. Things to look for: Initialized to values other than zero. Crashes and restarts without explanation, etc.

Images

Sample clip from top of web results at SD ROV.
(Just attach your images to this topic and they will automatically be added to the gallery above!)

Summary of articles submitted (Add | All):

Discussion List

See List Serve for all email discussion lists and to add more.

Project Form edit
Project Name Election Oversight, June 2016
Project Description Review of the primary election in California
Project Founder Ray Lutz
Project Curator Ray Lutz
Project Type Issue Oversight
Project Parents Election Integrity
Related Keywords Election Integrity, Snapshot Protocol, Snapshot Protocol
Project Status Hot
Thumbnail Link
Forum Link
List Serve Topic
I Attachment Action Size Date Who Comment
web_results_2016-06-09_5pm.jpgjpg web_results_2016-06-09_5pm.jpg manage 58.0 K 2016-06-12 - 00:50 Raymond Lutz Sample clip from top of web results at SD ROV.
web_results_2016-06-09_5pm.xmlxml web_results_2016-06-09_5pm.xml manage 59.8 K 2016-06-12 - 00:51 Raymond Lutz XML file showing canvass results.
This topic: Common > WebHome > ElectionOversightJune2016
Topic revision: 2016-08-29, RaymondLutz
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Copswiki? Send feedback