The letter came from a Republican. Al Schmidt, Pennsylvania's Secretary of the Commonwealth, the official who certified the state's elections through the angriest years anyone in his job has known, wrote to Homeland Security Secretary Markwayne Mullin on Wednesday with a question that a genuine mass-fraud discovery would never have to face: what data did you use? [1]
DHS claims "as many as 14,576 non-citizens registered to vote in Pennsylvania." Schmidt's letter notes, in his words, that "DHS has not identified its methodology or what data it relied on in reaching this conclusion," and asks the department to "provide a document explaining your methodology." [3] A Department of State spokesperson put it more plainly: "the federal government still has not shared the actual list of those people." [1]
Yesterday we reported the four-state picture: DHS letters announcing more than 250,000 suspected noncitizen registrations, against the two states that published their own numbers finding a combined 30,479 [callback]. The Pennsylvania response adds the piece that turns a discrepancy into a pattern, because the discrepancy is now inside DHS's own letters.
Here is the government's arithmetic, from the letters themselves. California: 190,832 claimed, 81,336 when matched on four fields. New Jersey: 35,152 claimed, 19,497 matched. Nevada: 15,903 claimed, 8,576 matched, a drop of 46 percent that state officials called "wildly speculative." Pennsylvania: 14,576 claimed, 8,594 matched. [2]
Add the public claims: 256,463. Add the four-field matches, the version where a name, a date of birth, an address and a Social Security number all have to line up: 118,003. The department's own stricter standard erases more than half of its own announcement. California alone accounts for a gap of 109,496 people.
There is a deeper problem, first flagged by Votebeat's reporting: the DHS letters say the department ran its comparison against "each state's public data files." Public voter rolls do not contain Social Security numbers. A four-field match that includes SSN data cannot be built from the files DHS says it used. [2] Either the matching was not what the letters describe, or the data was not what the letters describe. Schmidt's letter also points at the tool likely underneath it all, the SAVE database, whose modified version a federal judge recently found unlawful in several respects. [3]
The steel-man deserves stating. Voter rolls do contain errors, list maintenance is genuinely hard, and if even the matched numbers held up, thousands of improper registrations would be worth investigating. Schmidt himself has spent years saying exactly that, alongside the finding he repeated this week: "all evidence has shown that noncitizen voting is extremely rare across the country." [2] The way to investigate is the way he proposed, which is to hand the state the list and the method.
DHS has not responded to Schmidt's letter as of publication. Until it does, the record stands where the department's own documents put it: a public claim of 250,000, an internal match count of 118,003, a matching method its stated data cannot support, and a list that the officials responsible for the rolls have never seen. [1][2][3]
