Trump Campaign Lawsuit Regarding Maricopa Overvotes (and Sharpies)

Mick West

Administrator
Staff member
Article:
SUMMARY OF THE CASE

1. Qualified electors casting ballots in person on Election Day in Maricopa County submitted their completed ballot to an electronic tabulation machine. Numerous voters were alerted by these devices to a facial irregularity in their ballot—frequently an ostensible “overvote”—but were induced by poll workers to override the tabulator’s rejection of the ballot in the good faith belief that their vote would be duly registered and tabulated. In actuality, overriding the electronic tabulator’s alert automatically disqualifies the putative “overvotes” without additional review or adjudication.


Quick summary: some voters say (and some observers agree) that their in-person votes were rejected by the machine, and instead of being given an opportunity to review it, a poll-worker "pressed the green button" to accept the ballot, which meant that their vote in one or more of the things on the ballot, would have been marked as an "overvote".

Unofficial results actually list the number of overvotes.
https://recorder.maricopa.gov/media/Summary_110320.pdf

2020-11-07_15-49-11.jpg

That's 4,816 overvotes. You'd normally get some overvotes anyway, as people make mistakes. But how many? How many were there in previous years?

Past Maricopa results can be found here:
https://recorder.maricopa.gov/electionresults/archivedelectionresults.aspx

2016 Presidential Election:
https://recorder.maricopa.gov/electionarchives/2016/11-08-2016 Final Summary Report NOV 2016.pdf
2020-11-07_15-52-14.jpg

2020 had 8,209 undervotes, and 4,816 overvotes out of 2,052,756 votes. (overvote % =
2016 had 19,256 undervotes and 21,785 overvotes out of 1,608,875 votes! That's far more in 2020.
2012 had 6,648 undervotes, 3,229 overvotes out of 1,390,836 votes
2008 had 5,541 undervotes, 3,660 overvotes, out of 1,380,571 votes
2004 had 12,868 undervotes, 4,134 overvotes, out over 1,211,963

Arranging this as a table and calculating the percentage (as a percentage of votes cast)

YearUndervotesOvervotesVotes castUnder % Over %
20208209481620527560.40%0.23%
2016192562178516088751.20%1.35%
20126648322913908360.48%0.23%
20085541366013805710.40%0.27%
200412868413412119631.06%0.34%

Conclusion: the amount of undervotes and overvotes is consistent with previous years when taken as a percentage of total votes cast. It's actually LOWER than three of the previous four years. So this is not consistent with the idea that more ballots were marked as overvotes this year and suggests the broader claim of the lawsuit is without merit.

In addition, if the poll workers WERE erroneously letting overvotes pass through, they would probably have applied randomly to both candidates. So fixing this problem would result in a slight improvement for Biden, as he won Maricopa County.
 

Attachments

  • Maricopa County Undervotes and Overvotes as a percentage of votes cast. (1).xlsx
    5.4 KB · Views: 887
Last edited:
Conclusion: the amount of undervotes and overvotes is consistent with previous years when taken as a percentage of total votes cast. It's actually LOWER than three of the previous four years. So this is not consistent with the idea that more ballots were marked as overvotes this year and suggests the broader claim of the lawsuit is without merit.

am i missing something? i didnt read your lawsuit link so i'm going by what you posted there. where is a claim that more ballots were marked as overvotes?
the claim is that they were induced by poll workers to push a button that cancelled their vote.

did they push that button in previous years?

and is that last line in your quote summary accurate? or not?
 
am i missing something? i didnt read your lawsuit link so i'm going by what you posted there. where is a claim that more ballots were marked as overvotes?
the claim is that they were induced by poll workers to push a button that cancelled their vote.

did they push that button in previous years?
Pushing the button does not cancel the vote, it allows the ballot to go through with under/overvotes. It will only not count the vote items that seem to have more than one local marked. If it's a smudge, then only maybe 1 in 10 of them would be the Presidential vote. This would be reflected in the overvote total.

The lawsuit continues
Article:
2. Arizona law requires that putative overvotes be subjected to further review in an effort to discern the actual intent of the voter. While this safeguard was afforded to putative overvotes cast on early ballots and on Election Day ballots that poll workers properly segregated in a separate repository, potentially thousands of voters across Maricopa County have been disenfranchised by systematic improper tabulator overrides.


If there were thousands of people having their overvotes passed through instead of caught and corrected, then you'd expect much higher than normal overvotes. In fact, it's basically as low as it's ever been.

If there's a "systematic improper tabulator overrides" then it should affect the candidates equally.

and is that last line in your quote summary accurate? or not?
Can you clarify what you are asking about, and why you think it might not be accurate?
 
From the complaint:
DEMAND FOR RELIEF
WHEREFORE, the Plaintiffs demand relief in the following forms: [..]
B. [..] the Maricopa County Recorder and Maricopa County Board of Supervisors must:
(i) Identify (by means of electronic tabulation devices or otherwise) all Election Day ballots that contain apparent overvotes or other putative defects or irregularities in connection with the voter’s selection of a candidate that have not been adjudicated by the Electronic Vote Adjudication Board or the Ballot Duplication Board; [..]
C. [..] (i) shall not canvass or certify any returns in the November 3, 2020 general election unless and until the Maricopa County Recorder and Maricopa County Board of Supervisors have completed the process set forth in paragraph B above; [..]
So, basically, what they're hoping to achieve is a recount (aka scan all ballots again), and for ALL votes to be thrown out if the recount is not achieved (and for that to happen, all they'd need to do is slow it down until time runs out); presumably, in that case, the Republican-led Arizona legislature would determine the state's electors.

The court will probably investigate how these machines really work. I can't imagine them being designed to have a mechanism to "spit out" invalid ballots, but accepting overvotes via a key press. I can imagine them accepting undervotes; i.e. if someone didn't vote in a local race or other, then the machine might flag that as an undervote, and pressing the green button would let the vote be cast without those races that the voter didn't mark. (You might wish that legislation on how these machines work wasn't blocked in the US Senate, but I digress.)

What we do know is that the Sharpies are not the problem:
Article:
Arizona election officials confirmed that Sharpies were used in voting, but they said that would not invalidate a ballot. The Maricopa County Elections Department tweeted on Election Day that voting centers use Sharpies so that ink does not smudge when ballots are counted.

“New offset columns on the ballots means bleed through won’t impact your vote!” they tweeted in an informational video.

The officals selected them and presumably tested them, almost all voters used them, and most ballots were counted ok.

Mick's statistics are not meaningful if these "green-button" ballots weren't tallied as overvotes or undervotes; and if they're undervotes or overvotes for races other than the presidential election, they wouldn't be tallied at all in any case.
 
Last edited:
The court will probably investigate how these machines really work. I can't imagine them being designed to have a mechanism to "spit out" invalid ballots, but accepting overvotes via a key press. I can imagine them accepting undervotes; i.e. if someone didn't vote in a local race or other, then the machine might flag that as an undervote, and pressing the green button would let the vote be cast without those races that the voter didn't mark.
That sounds plausible. If you look at the Senate race on the same ballot, we have:
2020-11-07_23-45-39.jpg

And in 2012:
2020-11-07_23-46-26.jpg

A very large number of undervotes, as people presumably care less about their State Senator than about the US President.

Regarding "Press the Green Button" it seems like this would be accompanied with an explanation on-screen? Surely SOMEONE read this?

Although... (from the Lawsuit, page 23)

2020-11-07_23-51-25.jpg

I hope that there will be some statement from Maricopa soon.
 

Attachments

  • 2020-11-08_00-10-02.jpg
    2020-11-08_00-10-02.jpg
    162.1 KB · Views: 367
  • 2020-11-08_00-11-46.jpg
    2020-11-08_00-11-46.jpg
    302.6 KB · Views: 396
I've removed the word "Rebuttal" from the thread title, as I don't think it really rebuts the lawsuit.
 
"If the ballot has not been cast successfully, the ballot is likely blank or overvoted."
I'm reading this to imply that undervoted ballots don't require confirmation unless they're completely blank/unreadable.

But the documentation also says that "the screen [..] will have more information", which contradicts that affidavit. Hmm.
 
This video discusses how the machines operate:

Source: https://www.youtube.com/watch?v=yIuxo2RBXt8


This video shows two people demonstrating election-day voting using the machines.

Source: https://www.youtube.com/watch?v=eCc9AaN-xRc


Note these specific machines are ONLY used for the election day votes. So the percentage of possible errors is even lower. Mail-in and early in-person votes are collected and then tabulated using a different machine that counts in batches.

Close-up


The machine is made by Dominion voting. The small screen on the right provides feedback, with a large checkmark if the ballot is accepted. The process for a rejected ballot is not show, but according to the description above:
The screen will alert the voter whether or not the ballot has been cast successfully.

If the ballot has not been cast successfully, the ballot is likely blank or overvoted. If this is the case, the buttons below the screen (Cast-Green Oval, Return-Red Rectangle) will light up and the Marshal will alert the voter that they have a decision to make and refer them to the screen, which will have more information. The voter may need to use arrow buttons to scroll down to read all of the information. From there, the voter may decide to press the Return button to either mark their ballot or spoil their ballot and receive a new one, or press the Cast button to cast their ballot as is.
Content from External Source
 

Attachments

  • Maricopa County Election Day Voting Demo Dominion Machine.mp4
    677.8 KB
From the video, the "Green Button" works in the following cases:
  • Over Vote
  • Blank Ballot
The ballot is automatically returned for
  • Misread Ballot
  • Ambiguous Marks
Undervotes are not addressed, probably because they are so common that a partial undervote (where you vote for some contests but not all) is just let thought. Ballots often have contests on that people don't really care much about (like Judges, or obscure ballot measures) so they just leave them blank. I've done that.

In all cases, the screen gives the reason for the rejection. In the case of OverVotes it will tell you which contest it was, with contest 1 being the President.

Based on the video, it seems like you would only be able to press the red or green buttons after going through any additional pages of information about the error.
 
Article:
“There were 155,860 votes voted in person on Election Day (in Maricopa County),” Liddy continued. “Of those, the tabulator only identified 180 potential overvotes on the presidential line … 180, that’s it.”

And he said it would be “absurd” to assume that all 180 were incorrectly deemed overvotes.

“There is no possibility of systematic error with only 180 out of 155,860,” he said.

Looking at the data from Mick's first post:
2020-11-07_15-49-11.jpg
The overall overvote rate is 4816/2052756 = 0.23%.
For in-person voting, the rate is 180/155860 = 0.12%.
Obviously, when the vote tabulator can show voters that there's a problem, mistakes on in-person ballots can get corrected that wouldn't get corrected on absentee ballots, so that makes sense.

If there was a problem, it's small, and probably wasn't caused by the pens.
 
Last edited:
Conclusion: the amount of undervotes and overvotes is consistent with previous years when taken as a percentage of total votes cast. It's actually LOWER than three of the previous four years. So this is not consistent with the idea that more ballots were marked as overvotes this year and suggests the broader claim of the lawsuit is without merit.

In addition, if the poll workers WERE erroneously letting overvotes pass through, they would probably have applied randomly to both candidates. So fixing this problem would result in a slight improvement for Biden, as he won Maricopa County.
Hi Mick, thanks for your post.

I conducted an analysis of the overvotes and undervotes in Maricopa County, AZ, and found that there exists statistically significant correlation between Biden's performance and the percent of ballots found to be overvotes or undervotes:

maricopa_overundervotes_vs_biden_percent_per_precinct_2020.png

We can observe in this chart a positive correlation between Biden's performance in the precincts of Maricopa, AZ and the percent of ballots determined to have overvote or undervote errors in those precincts.
In other words, a higher rate of errors found in ballots in Maricopa precincts equals Biden having a higher share of votes.
We can see that in precincts where Trump had more than 50% of the vote share, the average error percentage in ballots was 0.62%, whereas it was 0.76% for Biden's precincts.
We can also see that on average, in cases where there was an error rate in ballots of over 1.2%, Biden had a 17% higher vote share than he did in precincts with lower rates of "error" ballots.
Most of the precincts in Maricopa with ballot error rates +1 standard deviation and greater from the mean were dominated by Biden.

Given these observations, there are two possible conclusions that can be drawn:
- Biden voters are, on average, less capable of properly filling out ballots than Trump voters.
- Fraud favoring Biden occurred more often or more heavily in precincts with higher rates of error ballots.

I do not think that Biden voters are less capable of correctly filling out ballots than Trump voters, so I would conclude fraud. Specifically, I would allege that in certain vulnerable precincts, higher rates of ballots that were meant for Trump are instead cast aside as overvotes or undervotes, yielding higher overall error rates and better performance for Biden in those precincts.

We can see that this differs from in 2016, where there still exists positive correlation between Clinton's performance and the percent of ballots that are overvotes and undervotes, but the strength of that correlation in 2016 is far less than it was this year, despite there being a higher percentage of overvotes and undervotes in 2016. Furthermore, the distribution of precincts with higher rates of undervotes and overvotes is visibly more uniform in 2016 than it is for 2020:
maricopa_overundervotes_vs_clinton_percent_per_precinct_2016.png

We can assume that errors *should* occur equally in both directions. With this said, we need an explanation for why precincts in Maricopa that have higher rates of overvotes and undervotes tend to favor Biden. Without a valid explanation for this, I can only conclude fraud.

Data sources:
https://recorder.maricopa.gov/electionarchives/2016/11-08-2016 Final Precinct Report NOV 2016.txt
https://recorder.maricopa.gov/media/ArizonaExportByPrecinct_110320.txt
 
Last edited:
Back
Top