California has 760,000 people who identify as Native American. More than any other state.

Its schools count 24,822 Native American students.

The actual number may be 156,000.

That’s a 90 percent undercount. Not a rounding error. Not a data gap. Ninety percent of Native American students in the largest state in the country, invisible in the system that’s supposed to serve them.

The mechanism is a checkbox.

California’s school enrollment forms ask students to identify their race and ethnicity. If a student checks both Hispanic and Native American, the system counts them as Hispanic. If they check Native American and any other race β€” Black, white, Asian β€” the system counts them as “two or more races.” In both cases, the Native American identity disappears. Not rejected. Overridden.

The form doesn’t say “pick one.” It lets you check every box. The database picks for you afterward.

Celestina Castillo, a Los Angeles parent and college learning center director, put it plainly: “If someone is Black, or Asian, or white, they’re counted that way. Why does it not count if someone is Native American? That’s not OK. It feels like erasure.”

It feels like erasure because it is erasure. It’s just erasure with a flowchart.

The technical name is the “Hispanic topcode.” Federal data collection standards, codified in OMB Statistical Policy Directive No. 15, historically treated Hispanic/Latino as an ethnicity that overrides racial identification. If you are Hispanic and anything else, you are Hispanic. The racial identity β€” whatever it was β€” gets absorbed. For most groups, this is a data inconvenience. For Native Americans, who are multiracial at a higher rate than any other major group, it is a disappearing act.

Brookings found the same pattern in higher education. As many as one in twenty undergraduate students identify as Native American alone or in combination β€” five times higher than the exclusive count. Enrollment has dropped 37 percent since 2010, more than twice the national average. But how much of that is students leaving, and how much is students being reclassified out of existence? The data can’t tell you. The data is the problem.

Here is what disappearing from the data means in practice. Schools with higher Native American enrollment qualify for federal and state funding for cultural services, tutoring, and curriculum centered on Native history. When 131,000 students are classified as something other than what they are, the funding follows the classification. The services follow the funding. The support follows the services. The kid who checked two boxes and got sorted into the wrong column doesn’t get any of it.

The U.S. Census Bureau solved this problem in 1970. Fifty-six years ago. They changed the methodology to allow people to write in tribal affiliations and identify as multiracial while still being counted as Native American. The result was visible immediately: in 1960, Native Americans were 0.3 percent of the U.S. population. By 2020, nearly 3 percent. The people didn’t multiply tenfold. The counting got honest.

Schools never caught up.

In March 2024, OMB finally revised Directive No. 15. The new standards remove the Hispanic override, require a combined race-and-ethnicity question that treats all categories equally, and replace “two or more races” with “multiracial and/or multiethnic.” Full compliance deadline: March 2029. Three more years. Three more graduating classes sorted into the wrong column.

California isn’t waiting. Assemblymember James Ramos β€” a member of the Serrano/Cahuilla tribe, the first California Native American elected to the state legislature in over a century β€” authored Assembly Bill 1581. The bill would let students write in their tribal affiliation on school forms and be counted as Native American regardless of what other boxes they check. No opposition has been reported.

“Native American students should be able to stand up in the classroom and say who they are and be proud of it,” Ramos told CalMatters.

Shannon Rivers of the California Native Vote Project, a member of the Akimel O’odham tribe, added something the data can’t capture: when Native American students are invisible in the count, it reinforces the stereotype that Native Americans are historical figures, not contemporary people. Not doctors, engineers, parents, students. Artifacts. The miscount doesn’t just lose funding. It loses the present tense.

There is a number that sits underneath all of this and doesn’t move.

Approximately 90 percent of California’s Native American population was killed β€” by violence, by disease, by policy β€” in the late 19th and early 20th centuries. Boarding schools took the children who survived and stripped their languages, their names, their practices. The explicit goal, in the words of the era, was to “kill the Indian, save the man.”

That was the first 90 percent.

The checkbox is the second.

The mechanism changed. The percentage held. In the 1800s, the disappearance was physical. In the 2020s, it’s administrative. The outcome is the same: you are no longer in the count. You are no longer in the funding formula. You are no longer in the room when decisions are made about the room.

A bill shouldn’t be necessary to count a child. A bill shouldn’t be necessary to let a student write the name of their nation on a form. A bill shouldn’t be necessary to stop a database from overriding an identity that predates the database, the state, and the country that built both.

But here we are. AB 1581. No opposition. Because even the system knows the system is wrong. It just needed someone to write the bill.

The override is not a glitch. It’s an architecture. And it has been running, uncorrected, for decades β€” while the Census Bureau that feeds the same government fixed the same problem fifty-six years ago.

The checkbox asks who you are. The database decides you’re wrong.

// NEON BLOOD

Sources: CalMatters β€” Carolyn Jones (May 21, 2026) Β· Brookings β€” Native American college student undercounting Β· Federal Register β€” OMB SPD 15 revision (March 2024) Β· California Assembly β€” AB 1581 analysis