The Arizona Canvass
A canvassing effort in Arizona done by Liz Harris was published recently. You can download or read the PDF by clicking below:
Some key highlights are listed below.
What the canvass was and how it got started:
What started small grew into a countywide grassroots effort drawing hundreds of volunteers from within the state and from across the country who collectively spent thousands of hours visiting the residences of almost 12,000 registered voters in Maricopa County. Volunteer programmers and database administrators joined together to database the voter data and build a mobile app for management and tracking of the canvass and the data it produced.
During the canvass, volunteers would visit the homes of registered Maricopa County voters. At the door, if the resident answered, the volunteers would identify themselves as private citizens conducting voluntary election integrity research and ask if they would mind answering a few questions.
They went out and directly got the data firsthand. It’s important to remember that this was not a forensic audit, it was a canvass.
Maricopa County comprises approximately 2,595,272 registered voters, making it the second largest voting district in the country and the third largest county by population. The canvass as conducted generally yields a confidence interval of 1.5% at a confidence level of 95%. Said another way, very similar to voting polls, the accuracy of these results applied to the county as a whole are within +/- 1.5% with 95% certainty, which is the scientific statistical standard for samples such as this. The results from the canvass are thus able to be applied with scientific certainty to the entire county as a whole.
Extent of canvass:
The “lost votes” problem:
The most problematic issue we have seen in the canvass comes from our analysis of the data showing that Maricopa County is missing votes from an estimated 173,104 voters.
During the canvass, hundreds of people who showed in the Maricopa County records as not voting in the election reported that they actually did vote in the election.
Interestingly, this data parallels the findings of a study conducted by Matt Braynard in November of 2020, where he phone surveyed 710 registered republican voters in Arizona who did not have a vote recorded by the State. In his survey 356, or 50.1%, of those surveyed stated that they had in fact cast a mail-in ballot.
The “ghost votes” problem:
The second-most problematic issue we have seen in the canvass comes from our analysis of the data showing that Maricopa County recorded an estimated 96,389 mail-in votes that likely could not have been physically cast by the voter that the vote was registered to. These voters did not have a secondary mailing address and were either unknown to the residents who lived at their voting address since September 2020 or were known but were confirmed to not have lived at the residence since prior to the election, and often had not lived there for many years. By law mail-in ballots are not forwarded, so it would not have been possible for these voters to have been in physical possession of their ballots.
A specific example to better illustrate what this problem looks like: One of the individuals that we canvassed has owned and lived at their home for more than a decade. For a time during 2010, they rented a room out to someone who later moved out- of Arizona. Looking back historically, although the renter had moved out of state, a vote had been cast under their name, by mail, continuing to use the same residence address, in the 2012, 2014, 2016, 2018 and 2020 elections. The homeowner is innocent in this. The person who moved away is innocent as well. But somehow another person or group of people has been able to fraudulently submit mail-in votes using the former renter’s information in multiple elections.
They also found other issues, such as:
Mail-in Voters Reported Actually Voting In-Person
Mail-in Voters Reported They Didn’t Actually Vote
Votes Cast Under Registrations Listing a Vacant Lot as their Address
The total error rate for only the “other issues” ballots is even high:
Added together, these issues impacted 5.18% of [all] mail-in votes
Executive summary wrapping it all up:
Two primary categories of ineligible ballots and election mishandling, encompassing hundreds of thousands of votes, render the 2020 General Election in Maricopa County uncertifiable.
An estimated 173,104 votes are missing or lost, as reported to our volunteers who went door to door verifying registration and voting information for thousands of residents. These are American citizens living in Maricopa County who cast a vote, primarily by mail, in the election and yet there is no record of their vote with the county and it was not counted in the reported vote totals for the election. Additionally an estimated 96,389 mail-in votes were cast under the names of registered voters who were either unknown to the residents of the registration address or who were verified as having moved away prior to October 2020. Other irregularities were uncovered during the canvass at a smaller scale, including votes cast by mail from vacant lots, votes recorded from residents who had not actually voted, etc.
These results are a travesty to our democracy and our voting rights. In addition to impacted local races, such as Maricopa County Board of Supervisors District 1, decided by 403 votes, key statewide race margins are well within the numbers shown above. The Presidential race was decided by 10,457 votes statewide, and the U.S. Senate election was decided by 78,886 votes statewide.
With this in mind, there remains a few notable problems with this canvassing project. I’m proud of those who worked on it and I’m sure it is indicative of some aspects of the fraud, but the report is lacking in some key areas. I do not like spreading false ideas, so we have to address the report issues. First, the address on the cover page appears to be an actual house when I look it up, so we do need some clarification on what that is supposing to show (edit: a reader mentioned that the correct address was supposed to be “2058 E Wildermuth, Tempe” which does show an empty lot in Google per March 2021, and that the report had been updated but my cover photo version is an improper version. Oh well, content is the same). Second, they don’t seem to mention anything about statistical bias analysis. They went to around ~12k homes and only got data on about 4.5k. There’s potential bias here especially in regards to the extrapolation and we don’t know how they addressed it. It’s important that we understand how they adjusted for this or they really shouldn’t be using extrapolated numbers at all, only percentages of the total 4.5k they could reach.
Still, let’s get this straight: Hundreds of thousands of votes were off in Arizona at a minimum (this canvass isn’t even including potential vote flipping or other nefarious actions which surely would make up the difference). Yet, the presidential election was decided by about 10k votes.
That seems safe and secure.
Yup, nothing to see here.