What a 40% Mismatch Reveals About the DOJ Noncitizen List and California’s Voter Rolls

A Department of Justice list of individuals alleged to be noncitizens has collided with California's voter registration records, and the collision produced a result that demands scrutiny. An analysis reported by AP News found that more than 40% of the names on that federal list do not appear on California's voter rolls at all.

The finding matters because the list's implicit purpose — identifying noncitizens who may have registered or voted — depends entirely on the accuracy of its matching. When nearly half the names fail to connect to any state registration record, the list's evidentiary value weakens considerably.

The reporting is notably thin on methodology, and that gap is itself part of the story. The report does not disclose how many names the list contains, how the comparison was performed, or who conducted the analysis.

It does not state whether the listed individuals were alleged to have registered, to have voted, or merely to have been flagged for review. Neither the Justice Department nor California election officials commented on the finding.

These omissions do not invalidate the analysis, but they limit how far any conclusion can responsibly be pushed.

What remains is a structural question about data integrity in election administration. A federal list that misidentifies more than four in ten names as state-registered voters is either overinclusive by design, built from unreliable source data, or matched against the wrong records. Each explanation carries different legal and political consequences.

This analysis examines the finding, the matching problem beneath it, and the governance questions that follow when federal and state data systems are asked to speak to one another.

TL;DR An AP News-reported analysis found that over 40% of names on a Department of Justice noncitizen list do not appear on California's voter rolls, raising direct questions about the list's accuracy and intended use. The report omits the list's size, the matching methodology, and whether individuals were alleged to have registered or voted. Neither DOJ nor California election officials responded publicly. The episode illustrates a recurring problem in cross-jurisdictional data matching: high error rates are structurally predictable when identifiers are weak, and the burden of proof falls on whoever publishes the list.

The Finding and What It Actually Establishes

The core number is straightforward: more than 40% of names on the DOJ list failed to match California voter registration records. That is a mismatch rate, not a finding of innocence or guilt for any individual named.

A name absent from the rolls could mean the person never registered, registered under a variant spelling, moved out of state, or was simply misidentified by the federal compilation. The analysis establishes a discrepancy; it does not establish why the discrepancy exists.

Reading the 40% Figure Correctly

Mismatch rates in record linkage are sensitive to the identifiers used. If the DOJ list relied on name-plus-date-of-birth matching without a state voter ID, common names and transposed dates will generate false positives at scale.

California's rolls contain roughly 22 million active registrations, so even a small per-record error rate produces thousands of spurious matches. The 40% figure is best read as a signal about list construction quality, not as a measure of noncitizen voting.

The report's silence on list size compounds the ambiguity. A 40% mismatch on a list of 200 names is a data-quality anecdote; the same rate on a list of 20,000 is a systemic failure.

Without the denominator, the percentage cannot be translated into an absolute count of affected individuals, and the practical harm — whether anyone faced removal proceedings or voter challenges — remains unquantified.

Equally unresolved is what the list was for. The report does not state whether names were alleged to have registered, to have voted, or merely to have been flagged for administrative review. Those are materially different allegations with different legal thresholds.

Conflating them in public discussion is precisely the kind of compression that turns a data artifact into a political claim.

Why the Silence From Both Sides Matters

Neither the Justice Department nor California election officials responded to the finding. Institutional silence in a dispute over voter data is not neutral; it leaves the public record shaped by whichever interpretation circulates fastest. DOJ could clarify the list's provenance and matching rules. California could confirm whether any listed individual was actually registered. Neither has done so publicly.

This vacuum has a predictable effect on credibility. When a federal agency publishes a list implicating individuals in a legally serious matter and declines to explain its construction, and when state officials decline to confirm or deny the matches, both institutions cede the interpretive field.

The result is a claim that can be asserted but not adjudicated, which is the worst outcome for public confidence in election integrity.

The Legal Standard the List Must Meet

Federal law permits scrutiny of voter list maintenance, but it also constrains it. The National Voter Registration Act bars systematic removal of registrants within 90 days of a federal election, a provision designed to prevent exactly the kind of late-cycle purges that list-driven challenges can trigger. A list with a 40% mismatch rate, deployed near an election, would test that protection directly.

Due process adds a second constraint. Individuals named on a federal list implicating them in a citizenship violation have an interest in notice and an opportunity to correct errors. The report does not indicate whether any such notice was given or whether any name was flagged in error. A list that circulates without a correction mechanism is, functionally, an accusation without an appeal.

The Data-Matching Problem Beneath the Headline

Record linkage across federal and state systems is a solved problem only in theory. In practice, it depends on shared identifiers, standardized name parsing, and tolerance thresholds that trade false positives against false negatives. When those elements are absent or inconsistent, mismatch rates climb predictably.

The California finding is less an anomaly than an illustration of what happens when linkage is attempted without them.

Why Cross-System Matching Fails at Scale

Voter files and federal databases were built for different purposes and rarely share a common key. California's rolls use county-level registration identifiers; federal compilations may rely on immigration records, tax data, or name-and-address strings. Absent a unique identifier, matching defaults to probabilistic name comparison, where common surnames and hyphenated names generate collisions. Each collision is a potential false positive.

The mathematics is unforgiving. If a matching rule has a 1% false-positive rate and is applied to a list of 10,000 names, roughly 100 individuals are wrongly implicated before any human review occurs. At a 5% rate, the figure reaches 500.

The observed 40% mismatch is consistent with a rule set that is either very loose or applied to source data of poor quality.

Where ##p## is the per-field error probability and ##n## the number of compared fields, the expression shows why adding fields without cleaning them can degrade rather than improve accuracy. A list built on three noisy fields can perform worse than one built on a single reliable identifier. The California result suggests the DOJ compilation did not resolve this trade-off.

What a Defensible List Would Require

A list intended to support enforcement would need documented provenance, a stated matching rule, a human review layer, and a correction pathway. It would also need a denominator and a published error estimate. None of these elements appears in the reporting.

Their absence does not prove bad faith, but it does mean the list cannot yet bear the weight placed on it.

Comparable federal data products — from tax administration to benefits eligibility — routinely publish methodology notes and error rates precisely because downstream users need to calibrate reliance. Election data deserves the same discipline.

A list that cannot state its own error rate is a list whose users cannot know how much of it to believe.

The Difference Between Mismatch and Misconduct

A name that fails to match a voter roll tells you one thing: the linkage failed. It does not tell you the person is a noncitizen, that they registered unlawfully, or that they voted. Those are separate factual questions requiring separate evidence.

Treating a mismatch as proof of misconduct inverts the burden of proof and converts an administrative artifact into an allegation.

This distinction is not pedantry. Voter challenges built on unmatched names have historically produced litigation and, in some cases, wrongful removal of eligible citizens from the rolls. The cost of a false positive falls on the individual, who must prove eligibility to regain a right they already held. That asymmetry argues for conservative thresholds before any list is used.

Governance, Timing, and the Political Economy of the List

The timing of the disclosure — tied to a midterm election cycle — is inseparable from its meaning. Voter-list disputes are rarely purely administrative; they arrive with electoral stakes attached.

A federal list with a documented 40% mismatch rate, released into a contested political environment, will be used by actors who have no interest in its error rate. The governance question is whether institutions can control that use.

Federal-State Tension Over Election Data

Election administration is constitutionally decentralized, and states guard their registration systems accordingly. Federal requests for voter data have repeatedly met resistance on privacy and preemption grounds. A DOJ list implicating California registrants sits at the intersection of federal enforcement authority and state administrative autonomy, an intersection with little settled law and considerable friction.

California's non-response is consistent with that friction. States have an interest in not validating federal compilations they did not build and cannot audit. The federal government has an interest in enforcement that does not depend on state cooperation.

When both sides withhold, the public receives a dispute without a resolution, and the underlying data problem persists unaddressed.

What Responsible Disclosure Would Look Like

A defensible process would publish the list's size, its matching criteria, and its estimated error rate before any name is used in enforcement or challenge. It would include a review layer where ambiguous matches are resolved by humans with access to additional records. It would notify individuals and provide a correction pathway with a defined timeline.

None of these steps is novel; they are standard practice in other domains where federal data drives adverse action against individuals. Their absence here is a choice, whether deliberate or by omission.

The remedy is procedural and unglamorous: document the method, publish the error rate, and let the list's accuracy determine how much weight it carries.

Reading the Episode Without Overreading It

The finding does not prove widespread noncitizen voting, and it does not prove the DOJ list was fabricated. It proves that a federal compilation and a state registration system disagreed on more than 40% of names.

That is a data-integrity finding with real consequences, and it deserves to be reported as such rather than absorbed into a pre-existing narrative.

The more durable lesson concerns institutional discipline. Lists that implicate individuals in legally serious conduct carry an obligation of accuracy proportional to their potential harm. When that obligation is unmet, the appropriate response is not to defend or attack the list wholesale, but to demand the methodology that would let anyone judge it.