Noncitizen Voting: What Republicans Claim vs. What the Data Shows

The Claim

  <div>
    <div>
      <p>The Republican Claim</p>
      <p>"3 to 5 million" illegal votes were cast in 2016, costing Trump the popular vote. Noncitizen voting is widespread, organized, and the reason Democrats win elections.</p>
    </div>
    <div>
      <p>What the Data Shows</p>
      <p>Across every state audit, federal investigation, and academic study, noncitizen voting occurs at rates between <strong>0.0001%</strong> and <strong>0.007%</strong> of votes cast. No evidence it has ever affected any election outcome.</p>
    </div>
  </div>

  <p>In January 2017, President Trump told congressional leaders that 3 to 5 million people voted illegally in 2016 — all for Hillary Clinton, none for him. He provided no evidence. PolitiFact rated the claim "Pants on Fire." FactCheck.org called it "bogus." No election official or academic expert supported it.<sup><a href="#s1">[1]</a></sup> <sup><a href="#s2">[2]</a></sup></p>

  <p>Trump then created the Presidential Advisory Commission on Election Integrity to prove the claim. Eight months later, he disbanded it — without releasing any evidence of widespread noncitizen voting. But the claim didn't die. It became the centerpiece of Republican election messaging through 2024 and beyond, used to justify the SAVE Act, voter purges, and proof-of-citizenship requirements.</p>

  <p>So let's look at what the evidence actually shows.</p>

  <h2>The Audits: What Republican Officials Actually Found</h2>

  <p>The most damning evidence against the noncitizen voting narrative comes from Republican election officials who went looking for it.</p>

  <table>
    <thead>
      <tr>
        <th>State</th>
        <th>Year</th>
        <th>Conducted By</th>
        <th>Registered Voters</th>
        <th>Noncitizens Found</th>
        <th>Actually Voted</th>
        <th>Rate</th>
      </tr>
    </thead>
    <tbody>
      <tr>
        <td><strong>Georgia</strong></td>
        <td>2024</td>
        <td>Brad Raffensperger (R)</td>
        <td>8,200,000</td>
        <td>20</td>
        <td>*</td>
        <td>0.00024%</td>
      </tr>
      <tr>
        <td><strong>Ohio</strong></td>
        <td>2024</td>
        <td>Frank LaRose (R)</td>
        <td>8,000,000+</td>
        <td>597</td>
        <td>138</td>
        <td>0.007%</td>
      </tr>
      <tr>
        <td><strong>Michigan</strong></td>
        <td>2025</td>
        <td>MI Dept. of State</td>
        <td>7,200,000</td>
        <td>16</td>
        <td>16</td>
        <td>0.00022%</td>
      </tr>
      <tr>
        <td><strong>Kansas</strong></td>
        <td>1999–2018</td>
        <td>Federal court (Fish v. Kobach)</td>
        <td>1,800,000</td>
        <td>39</td>
        <td>5</td>
        <td>0.002%</td>
      </tr>
      <tr>
        <td><strong>Pennsylvania</strong></td>
        <td>2017</td>
        <td>Al Schmidt (R)</td>
        <td>8,700,000</td>
        <td>220</td>
        <td>—</td>
        <td>0.0025%</td>
      </tr>
      <tr>
        <td><strong>Iowa</strong></td>
        <td>2024</td>
        <td>IA Sec. of State</td>
        <td>—</td>
        <td>277**</td>
        <td>35</td>
        <td>—</td>
      </tr>
      <tr>
        <td><strong>Utah</strong></td>
        <td>—</td>
        <td>UT review</td>
        <td>2,100,000</td>
        <td>1</td>
        <td>0</td>
        <td>0.00005%</td>
      </tr>
      <tr>
        <td><strong>Texas</strong></td>
        <td>2019</td>
        <td>David Whitley (R)</td>
        <td>16,000,000</td>
        <td colspan="3"><em>Claimed 95,000–100,000. Entire list rescinded after mass errors — mostly naturalized citizens. Whitley lost his job.</em></td>
      </tr>
    </tbody>
  </table>

  <p><small>* Georgia's Raffensperger reported 20 noncitizens found on rolls, all referred for investigation. A separate 2022 investigation found 1,634 noncitizens <em>attempted</em> to register 1997–2022, but all were blocked before registration was completed.<sup><a href="#s3">[3]</a></sup><br>** Iowa initially claimed 2,176, revised down to 277 (12.7% of original claim) after investigation.<sup><a href="#s4">[4]</a></sup></small></p>

  <p>Note the pattern: Republican officials go looking for noncitizen voters. They find a handful — dozens, occasionally hundreds — out of millions. In Georgia, Raffensperger's own COO said it plainly: <strong>"There is no proof that there is this overwhelming number of noncitizens on the rolls."</strong><sup><a href="#s3">[3]</a></sup></p>

  <p>In Pennsylvania, Republican Secretary Al Schmidt — who <em>discovered</em> noncitizen registrations caused by a PennDOT programming error — concluded: <strong>"One thing that became very clear through that research and all evidence suggests that noncitizens voting in elections in the United States occurs very rarely."</strong><sup><a href="#s5">[5]</a></sup></p>

  <p>And in Texas, the entire audit collapsed under its own errors. Secretary Whitley announced 95,000–100,000 potential noncitizens on voter rolls. Within weeks, tens of thousands were confirmed as naturalized citizens whose records predated their citizenship. The state was forced to rescind the entire list. Three lawsuits were settled. Whitley lost his position.<sup><a href="#s6">[6]</a></sup></p>

  <h2>Trump's Own Commission</h2>

  <p>In May 2017, Trump created the Presidential Advisory Commission on Election Integrity, chaired by VP Mike Pence and vice-chaired by Kansas Secretary of State Kris Kobach — the nation's most prominent advocate of the noncitizen voting narrative.</p>

  <p>The commission lasted eight months. It held two meetings. It was dissolved by executive order on January 3, 2018.</p>

  <div>
    <span>Disbanded without findings</span>
    <p>Presidential Advisory Commission on Election Integrity (2017–2018)</p>
    <p>The White House cited state resistance and legal battles. Commission member Matthew Dunlap (Maine Secretary of State, Democrat) contradicted Trump's claim of "substantial evidence," stating the commission found no such evidence. ProPublica obtained commission documents showing a draft report section titled "Evidence of Election Integrity and Voter Fraud Issues" was <strong>almost entirely blank</strong>.<sup><a href="#s7">[7]</a></sup> <sup><a href="#s8">[8]</a></sup></p>
    <p>The National Association of Secretaries of State issued a bipartisan statement: they were <strong>"unaware of any evidence that supports the voter fraud claims."</strong><sup><a href="#s9">[9]</a></sup></p>
  </div>

  <p>Mississippi's Republican Secretary of State Delbert Hosemann responded to the commission's data request by telling it to <strong>"go jump in the Gulf of Mexico."</strong><sup><a href="#s7">[7]</a></sup></p>

  <h2>The Day Voter Fraud Had Its Day in Court</h2>

  <p>The definitive test of the noncitizen voting claim happened in a federal courtroom in Kansas. Kris Kobach — the nation's leading proponent of the theory — had implemented a documentary proof-of-citizenship requirement for voter registration. The ACLU challenged it. Kobach got a full federal trial to present his evidence.</p>

  <div>
    <span>Case failed</span>
    <p>Fish v. Kobach (2018) — U.S. District Court, Kansas</p>
    <p>In a <strong>118-page ruling</strong>, Chief District Judge Julie Robinson struck down Kobach's proof-of-citizenship law and delivered one of the most quoted lines in election law:</p>
  </div>

  <blockquote>
    There is no iceberg; only an icicle, largely created by confusion and administrative error.
    <cite>— Judge Julie Robinson, Fish v. Kobach (2018)</cite>
  </blockquote>

  <p>The court found:<sup><a href="#s10">[10]</a></sup> <sup><a href="#s11">[11]</a></sup></p>
  <ul>
    <li>Over <strong>20 years</strong>, fewer than <strong>40 noncitizens</strong> had registered in a Kansas county with 130,000 voters</li>
    <li>Only <strong>5 of those 40</strong> ever cast a vote</li>
    <li>Most registrations resulted from <strong>confusion and administrative error</strong>, not deliberate fraud</li>
    <li>The court found "no credible evidence that a substantial number of noncitizens registered to vote"</li>
    <li>Meanwhile, the proof-of-citizenship law had blocked <strong>more than 35,000 Kansas citizens</strong> from registering</li>
  </ul>

  <div>
    <p>35,000 : 5</p>
    <p>Kansas citizens <strong>blocked from registering</strong> by the proof-of-citizenship law vs. noncitizens who <strong>actually voted</strong> over 20 years. Kobach was held in <strong>contempt of court</strong> for "willful failure" to comply with prior orders. The judge called his explanations "nonsensical" and "disingenuous."<sup><a href="#s11">[11]</a></sup></p>
  </div>

  <p>The 10th Circuit Court of Appeals unanimously upheld the ruling in 2020.</p>

  <h2>The Academic Record</h2>

  <p>One academic study is responsible for virtually every claim that noncitizen voting is widespread. It has been comprehensively rebutted — by the creators of the dataset it used.</p>

  <div>
    <span>Rebutted</span>
    <p>Richman, Chattha & Earnest (2014) — "Do non-citizens vote in U.S. elections?"</p>
    <p><strong>Published in:</strong> <em>Electoral Studies</em></p>
    <p><strong>Claimed:</strong> 6.4% of noncitizens voted in the 2008 presidential election. Extrapolated this to argue noncitizen votes could have been decisive in close races.</p>
    <p><strong>Method:</strong> Used survey data from the Cooperative Congressional Election Study (CCES), in which some respondents who identified as citizens in one survey wave later identified as noncitizens in a subsequent wave (and vice versa).</p>
    <p>This study was cited by Trump, Fox News, and Republican legislators as proof of widespread noncitizen voting.<sup><a href="#s12">[12]</a></sup></p>
  </div>

  <div>
    <span>Definitive rebuttal</span>
    <p>Ansolabehere, Luks & Schaffner (2015) — "The Perils of Cherry Picking Low Frequency Events"</p>
    <p><strong>Published in:</strong> <em>Electoral Studies</em></p>
    <p><strong>Authors:</strong> The <em>coordinators of the CCES itself</em> — the very dataset Richman used (Harvard, YouGov, UMass Amherst).</p>
    <p><strong>Finding:</strong> The CCES had a citizenship classification reliability rate of 99.9%, meaning 0.1% of respondents who were citizens accidentally clicked "noncitizen" in one wave. Applied to the large number of citizens in the survey, this tiny error rate produced <em>exactly</em> the number of "noncitizen voters" Richman reported. Panel data confirmed: the same individuals switched between "citizen" and "noncitizen" across survey waves.</p>
    <p><strong>Conclusion:</strong> "The likely percent of non-citizen voters in recent US elections is <strong>0</strong>."<sup><a href="#s13">[13]</a></sup></p>
  </div>

  <p>When Richman himself revisited the question using improved 2022 CCES methodology, he found just under 1% of noncitizens <em>indicated they had registered</em> — orders of magnitude below his original claim, and consistent with ongoing survey response error rather than actual voting.<sup><a href="#s12">[12]</a></sup></p>

  <p>The Brennan Center surveyed election officials in 42 jurisdictions covering <strong>23.5 million votes</strong> in the 2016 election. They found approximately <strong>30 suspected noncitizen voting incidents</strong> — a rate of <strong>0.0001%</strong>. Forty of 42 jurisdictions reported <strong>zero</strong>.<sup><a href="#s14">[14]</a></sup></p>

  <h2>The Heritage Foundation's Own Database</h2>

  <p>The Heritage Foundation — a conservative think tank that actively advocates for voter ID laws — maintains the most comprehensive database of documented election fraud in the United States.</p>

  <div>
    <span>Source confirms the opposite</span>
    <p>Heritage Election Fraud Database</p>
    <p>Total proven fraud cases of <em>all types</em> across all elections going back decades: approximately <strong>1,500</strong>.</p>
    <p>Noncitizen-specific cases: approximately <strong>77 instances</strong> of noncitizens successfully casting ballots between 1999 and 2023, per the Bipartisan Policy Center's analysis of the Heritage data.<sup><a href="#s15">[15]</a></sup> <sup><a href="#s16">[16]</a></sup></p>
    <p>Heritage itself describes the database as "a sampling of proven instances" and acknowledges it is "not comprehensive."</p>
  </div>

  <p>Seventy-seven cases over 24 years, across a country where <strong>billions of votes</strong> have been cast in that period. The Heritage Foundation's own data — the data compiled by the organization most motivated to find noncitizen voting — shows it is negligible.</p>

  <h2>Why Noncitizens Don't Vote</h2>

  <p>The rarity of noncitizen voting is not mysterious. The penalties are catastrophic:</p>

  <div>
    <div>
      <p>5yr</p>
      <p>Federal prison for registering to vote as a noncitizen</p>
    </div>
    <div>
      <p>DEPORT</p>
      <p>Deportable offense regardless of immigration status or years of residence</p>
    </div>
    <div>
      <p>BAN</p>
      <p>Permanently inadmissible to the U.S. — no visa, no green card, <strong>no waiver available</strong></p>
    </div>
    <div>
      <p>NO PATH</p>
      <p>Permanently barred from ever becoming a U.S. citizen</p>
    </div>
  </div>

  <p>A noncitizen who votes risks losing everything — their legal status, their home, their family's future in the country — for a single vote that will not change any election outcome. A long-term green card holder can be placed in removal proceedings over one vote.<sup><a href="#s17">[17]</a></sup></p>

  <p>Additionally, voter registration already requires attestation of citizenship <strong>under penalty of perjury</strong>. States cross-reference registrations against Social Security Administration, DHS, DMV, and interstate databases.<sup><a href="#s18">[18]</a></sup></p>

  <h2>The SAVE Tool: When the Cure Is Worse Than the Disease</h2>

  <p>In 2025, DHS expanded the Systematic Alien Verification for Entitlements (SAVE) database for voter eligibility checks. The results have been disastrous.</p>

  <div>
    <span>81% false positive rate</span>
    <p>SAVE Tool Voter Verification (2025–2026)</p>
    <p><strong>St. Louis County, Missouri:</strong> SAVE flagged <strong>691 voters</strong> as noncitizens. After investigation, 35% were immediately confirmed as naturalized citizens. After further cross-referencing, the list was cut to 133 — meaning <strong>at least 81% of the original flags were incorrect</strong>.<sup><a href="#s19">[19]</a></sup></p>
    <p><strong>70 county clerks</strong> across Missouri — both parties — sent a letter warning SAVE "repeatedly flags individuals we know to be U.S. citizens."</p>
    <p>Over 70 voting agencies across 24+ states use SAVE. The Texas Tribune reported DOGE "optimized" the database over two weeks, rushing it into use before it could discern voters' up-to-date citizenship information.<sup><a href="#s19">[19]</a></sup></p>
  </div>

  <p>The pattern repeats: the tools built to find noncitizen voters primarily flag <em>naturalized citizens</em> — legal Americans who went through the citizenship process — creating chaos without finding meaningful fraud.</p>

  <h2>When Cato Agrees with the Brennan Center</h2>

  <p>The Cato Institute is a libertarian think tank. It is not aligned with Democrats. It supports less immigration regulation, not more. And its analysis of noncitizen voting is unambiguous:</p>

  <blockquote>
    Noncitizens don't illegally vote in detectable numbers.
    <cite>— Alex Nowrasteh, Cato Institute</cite>
  </blockquote>

  <p>Cato has published multiple analyses calling Republican noncitizen voting claims "false" and "bogus," based on the same state audit data compiled above. Their Utah finding: out of 2.1 million registered voters, <strong>one</strong> confirmed noncitizen — who had <strong>never voted</strong>.<sup><a href="#s20">[20]</a></sup></p>

  <p>When the conservative Heritage Foundation's own database, the libertarian Cato Institute, Republican election officials in Georgia, Ohio, and Pennsylvania, Trump's own election commission, and a federal court all reach the same conclusion — the question is settled.</p>

  <h2>The Counter-Arguments</h2>

  <h3>1. "We can't find it because we don't look" (Score: 2/5)</h3>
  <p>Some argue the low detection rate reflects inadequate investigation, not a low actual rate — an unfalsifiable "absence of evidence isn't evidence of absence" claim.</p>
  <p><strong>Why it fails:</strong> Multiple states <em>have</em> aggressively looked. Texas spent significant resources and found nothing (and humiliated itself). Kobach got a federal trial specifically to present his evidence. Trump created a presidential commission. Republican secretaries of state in Georgia, Ohio, and Pennsylvania conducted targeted audits. Every investigation found the same thing: near-zero rates. This isn't a lack of investigation — it's a surplus of investigation producing the same null result.<sup><a href="#s3">[3]</a></sup> <sup><a href="#s6">[6]</a></sup> <sup><a href="#s10">[10]</a></sup></p>

  <h3>2. The Richman 2014 study (Score: 1/5)</h3>
  <p>The most-cited academic study claiming 6.4% of noncitizens voted.</p>
  <p><strong>Why it fails:</strong> The creators of the dataset Richman used published a peer-reviewed rebuttal demonstrating the finding was an artifact of survey response error. Richman himself found much smaller numbers with improved methodology. The academic consensus is clear: the 2014 study does not hold up.<sup><a href="#s13">[13]</a></sup></p>

  <h3>3. "Even a few noncitizen votes undermine election integrity" (Score: 3/5)</h3>
  <p>This is the strongest argument — not that noncitizen voting is widespread, but that <em>any</em> amount is unacceptable and warrants preventive measures.</p>
  <p><strong>Where it has merit:</strong> It's true that even one fraudulent vote is illegal and should be prosecuted. The question is proportionality: does the remedy cause more harm than the problem? In Kansas, proof-of-citizenship requirements blocked 35,000 citizens to prevent 5 noncitizen votes over 20 years. The SAVE tool flags citizens as noncitizens at an 81% false-positive rate. The cure is measurably worse than the disease, and the "election integrity" framing is used to justify restrictions that disproportionately burden eligible voters.<sup><a href="#s10">[10]</a></sup> <sup><a href="#s19">[19]</a></sup></p>

  <h2>Verdict</h2>

  <div>
    <h3>Assessment</h3>
    <p><strong>The claim that illegal immigrants vote for Democrats in meaningful numbers is false.</strong> It is not a matter of interpretation or political perspective. It is contradicted by:</p>
    <p>Every state-level audit conducted by Republican officials (Georgia, Ohio, Pennsylvania). Trump's own Presidential Advisory Commission on Election Integrity (disbanded without findings). A federal court trial where the nation's leading proponent of the theory presented his best evidence and lost (Fish v. Kobach). The Heritage Foundation's own fraud database (77 cases in 24 years). The Cato Institute's independent analysis. The Brennan Center's survey of 23.5 million votes (30 incidents, 0.0001%). And the definitive academic rebuttal by the creators of the very dataset the original claim rested on.</p>
    <p>Noncitizen voting exists. It is real. It is also vanishingly rare — measured in the dozens per state, not the millions Republicans claim. It has never been shown to affect any election outcome. The individuals involved typically registered through confusion or administrative error, not a deliberate scheme to swing elections.</p>
    <p>The political function of the claim is not to address a real problem. It is to justify voter restrictions — proof-of-citizenship laws, voter purges, registration barriers — that block far more eligible citizens than noncitizens. In Kansas, the ratio was 7,000 to 1. The claim is the pretext. The restrictions are the point.</p>
  </div>

  <h3>Adversarial Check</h3>
  <p><strong>Premise check:</strong> Is "do noncitizens vote" the right question? Partly. The more precise questions are: (1) at what rate, (2) does it affect outcomes, and (3) do proposed remedies cause more harm than the problem. The analysis above addresses all three.</p>
  <p><strong>Strongest counter-argument (scored 3/5):</strong> "Any amount of noncitizen voting is unacceptable." This has moral force but does not support the specific Republican claim of widespread, election-altering fraud. It also does not justify remedies that block thousands of citizens per noncitizen caught. The proportionality question is the dispositive one, and the data is clear: current proposals cause orders-of-magnitude more citizen disenfranchisement than the noncitizen voting they prevent.</p>
  <p><strong>No unforced concessions.</strong> The Richman study was included and addressed. The "any fraud is too much" argument was engaged at full strength. The verdict reflects the weight of evidence from across the political spectrum — not a partisan conclusion, but a factual one.</p>