ESII Study Concludes Maricopa Signature Verification Process Deeply Unreliable
March 9, 2022

CAMBRIDGE, MASS. – Today, the Elections System Integrity Institute (ESII) released the findings of a pilot and extended study into Maricopa County's "rigorous signature verification process."

For years, until it no longer served certain political agendas, there was bipartisan agreement that the signature verification process was flawed and insufficient. However, since the 2020 election, Maricopa County has doubled down on its assertions that the process is "rigorous" and has maintained its claim that of the 1,911,918 early mail ballots (EVBs) "verified by trained staff," only 25,000 were flagged as mismatches requiring review and only 587 (2.3%) of the cured were labeled "bad" signatures.

ESII sought to verify these claims in an initial pilot study followed by an extended study that is accessible below. Maricopa County denied requests for the ID signature images used in the matching process, so the ESII matched the signatures on envelope images with signatures from deeds.

The results vastly contradicted Maricopa County's assertions.

Both of ESII's studies used three experts (Forensic Document Examiners – FDEs) and three trained novices (non-FDEs trained with Maricopa County's signature verification manuals).

In the pilot study of 499 EVBs chosen randomly from the 1,911,918 (yielding a 95% confidence level and a margin of error of ±4.4%), ALL concurred that 60 of the 499 (12%) EVBs to be
signature mismatches. The pilot study concluded that 229,430 EVBs should have been cured versus the "upwards of 25,000" that Maricopa County reported cured.

In the extended study, 2,770 EVBs were chosen randomly from the 1,911,918 (yielding a 99% confidence level and a margin of error of ±2.5%). In an abundance of caution and to account for some who might argue that the deed signatures ESII matched against were wrong, ESII excluded EVBs that all six reviewers said were not matching. A minimum of 215,856 EVBs (11.3%) still had mismatched signatures that should have been cured.

"Simply put, these studies, no matter how you analyze them, require that we consider several questions," said Shiva Ayyadurai, Founder and Engineer at ESII. "First, if three experts and three non-experts all had varying signature cure rates, how can we count on this process being accurate? Second, if even 5%, let alone 10% of signatures need to be cured, that represents tens of thousands of ballots."

"Our current system of using signature verification for voting is broken. There is not reliability or consistency in application." Ayyadurai concluded. "The good news is, for the first time, we
now know how much this process if flawed."

ESII invites Maricopa County to provide the signature images they used in the curing process so that the experiment can be redone to mimic the exact process and data used by the County.
Our elections are too important to leave it up to guessing if two signatures match properly or not.

###

Phone: 1-617-631-6874
Email: Press@ElectionSystemsIntegrity.org