Proposition 39 Offers a Voter Confidence Rationale for Stricter ID Checks
Hoover’s Justin Grimmer argues that Proposition 39 is more likely to affect confidence in California elections than to prevent significant voter fraud, which he says is already rare. The measure would replace signature checks on mail ballots with a match against the last four digits of a government ID and require ID at polling places; Grimmer says research finds voter ID laws have little effect on turnout but may increase trust. He also identifies an unresolved implementation question: how voters already registered would choose an ID for mail-ballot verification.

The case for Proposition 39 is confidence more than fraud prevention
Justin Grimmer describes Proposition 39 as changing how California verifies voters, not as a response to widespread illegal voting. In his account, fraud in the United States—both in-person and by mail—is very rare, though not nonexistent. That leaves little room for an identification requirement to reduce it substantially. The measure’s stronger potential benefit, he says, is that clearer, more automatic identity checks could increase public confidence in elections.
That distinction matters because Grimmer sees two conditions coexisting: elections have high participation and are almost entirely free of fraud, while many people remain skeptical about election results and how elections are run. He points to research that he says finds voter trust increases after states pass voter ID laws. He expects that effect could be particularly relevant among Republicans who have expressed skepticism about California’s election administration. He also says voter ID laws are popular among Republicans, Democrats, and independents, and may create a perception that elections are more secure.
Voter ID laws could make people feel better about what's going on in their elections, and could lead to increased trust.
The proposal would change identity checks for both mail and in-person voting, but in different ways. For mail ballots, voters would provide the last four digits of the government-issued identification they designated for verification, replacing the current signature comparison. At the polls, voters now provide their name and address, along with a signature for a poll worker to confirm; if the measure passes, they would provide government-issued identification. Grimmer describes these as changes to existing safeguards, not the introduction of identity checks where none existed.
A separate provision would require counties and the state to make efforts to verify that registered voters are U.S. citizens. Local election administrators would also have to report on those efforts and on the share of voters potentially removed because they are not citizens. Grimmer describes a requirement for verification efforts and reporting, not a particular investigative method.
The current safeguards can be difficult for some voters to navigate
Signature matching is already part of California’s mail-ballot verification, but it can pose difficulties for voters who cannot reproduce the signature on file. Grimmer recalls a recent college graduate who had signed a line, rather than her usual signature, when registering at the DMV. When she later planned to vote by mail, she wondered how officials would verify her ballot if its signature did not resemble the one she had submitted.
The challenge is not limited to young voters. Grimmer says some people with disabilities cannot replicate their signatures, and older people may find it harder as their hands become less steady. A mismatch, then, does not necessarily indicate an attempt to cast someone else’s ballot; it can also reflect a voter’s difficulty reproducing a mark.
A mismatch triggers a further opportunity to verify identity. The voter is contacted and can provide identification or go to a local election office, among other ways of resolving the issue. But those steps require extra effort. If the voter cannot be reached or does not complete them, Grimmer says, the ballot is rejected and ultimately not counted.
The proposed replacement would use the last four digits of the government ID the voter selected. For someone who registered using a driver’s license, for example, that would mean providing the last four digits of the license number when returning a mail ballot. The source of the check changes: instead of an election official comparing two signatures, the ballot’s digits would be compared with the digits on file. Grimmer’s description is of one safeguard replacing another, rather than verification being added where none existed.
At a polling place, voters currently give their name and address, and a poll worker checks that they are on the voter records and confirms their identity with a signature. Grimmer says a voter who registers in person at the polling place must provide identification for the ballot to be cleared. He also notes that, under current practice, a voter who offers an ID at the polls as a way to verify identity may find that the poll worker is not permitted to take it. That can contribute to concerns about how identity is being checked.
The proposal would narrow some accepted IDs and formalize citizenship checks
For online voter registration, Grimmer says, California generally asks for a driver’s license number or Social Security number and uses that information to verify identity. Those government-issued identifiers would remain options under Proposition 39.
The change would affect people who lack those numbers or do not want to provide them. Under current law, they can use identification issued by a business in the normal course of business to verify their identity. Grimmer says this category can include a photo ID from a health club, an example that has drawn criticism. He emphasizes that very few voters have their identities confirmed this way.
The citizenship provision addresses a different stage of the process from the identity check when a person casts a ballot. Grimmer discusses allegations, including one from the Trump administration that 20,000 noncitizens were on the rolls in Los Angeles County or California. He says claims of this kind can be unreliable because matching people across different sources of information is difficult.
One recurring source of error, he says, is an outdated list of people with green cards. Someone on such a list may since have become a U.S. citizen and registered to vote, while a comparison against the old record still classifies that person as a noncitizen. In investigations of claims that large numbers of noncitizens are registered, Grimmer says, careful review has generally produced much smaller numbers—often a handful or perhaps a dozen.
He thinks requiring counties to investigate citizenship status could address concerns that the state has been too lax. But he does not expect careful investigation to confirm figures like the one he discusses; in his view, such numbers may reflect outdated data or the difficulty of matching records. The proposal’s reporting requirement would make counties’ efforts and the share of potentially removed voters visible, without specifying a single method for conducting those checks.
The turnout evidence does not settle the choice of verification standard
Critics of California’s signature procedure, as Grimmer describes their argument, say officials presume that signatures match and look for large, material differences. They view that as too passive a way to verify identity. Grimmer says his research has not found that the procedure invites widespread fraud. Still, he sees a difference in how the two systems work and may be perceived: comparing signatures requires judgment about whether two marks are sufficiently alike, while comparing digits requires checking whether the numbers on the ballot match those on file.
That more direct check, he says, could make the process seem clearer and increase confidence, even if it does little to reduce a level of fraud that is already low. The confidence rationale is therefore distinct from a claim that the current system permits extensive illegal voting.
The evidence on participation is different from the evidence on trust. Grimmer says research on voter identification laws finds essentially no effect on turnout, and that any effect on whether Republicans or Democrats have an advantage is smaller still. He argues that predictions that Proposition 39 would make voting much harder for large numbers of people and deter many from voting overstate the likely effects.
The measure would provide an opportunity for people without identification to obtain a state-issued ID. Grimmer acknowledges that doing so takes effort, as does voting more generally. The question for voters, in his framing, is whether that effort is worth a change in how elections verify identity—not whether voting should require no effort at all.
The choice, then, is about what standards voters want election law to embody. Some may prefer continuity with a signature process people know; others may value an automatic check with a clear standard for confirming identity, whether voting by mail or in person. Grimmer says the outcome is exceedingly unlikely to decide any California race.
Existing mail voters still need a way to choose an ID
The implementation question Grimmer identifies is how people already registered to vote would designate the identification they want to use for mail-ballot verification. He says the law is unclear on that point and identifies it as a challenge for the state legislature to clarify.
That designation matters because the proposed mail-ballot check depends on a voter providing the last four digits of the government ID selected for verification. For a voter who registers in the future, Grimmer says, the state already requests a driver’s license number or Social Security number during online registration, and that information is used to verify identity. But he does not describe how existing voters would select or record an ID for the new mail-ballot process. The required digits cannot be matched until that choice has been made and is available to election administrators.


