In Florida schools where almost all students are black or Hispanic, 13% of black students were classified as having a disability. Yet in schools where the vast majority of students were white, nearly 22% of black students get classified that way.

It’s a striking divide, and one that researchers say probably shouldn’t exist. The more accurate number is likely somewhere in between.

The result: Lots of black students may be going without services they need, and other black students are getting services they don’t — and potentially being pulled out of regular classrooms in the process.

Those are the findings of a new study looking at special education in the country’s third-biggest state, one that adds important new context to an ongoing debate about race and special education.

Civil rights groups and the Obama administration have backed controversial federal rules designed to address over-representation of some groups in special education — regulations that are being enacted now despite the Trump administration’s continued effort to delay them.

Meanwhile, researchers have been locked in a fraught debate over whether students of color are identified as disabled too often or not often enough. The latest research suggests the answer is both, depending on the racial makeup of the school. In other words: School segregation and special education are interconnected.

“The truth is complicated: Black and Hispanic students are over-identified with disabilities in predominantly white schools and substantially under-identified in schools with large shares of minority students,” said Claudia Persico, one of the Florida researchers.

Another new study found that students of color in Wisconsin were much more likely to be identified as emotionally disturbed or intellectually disabled if they attended a predominantly white school.

“These two studies cut through the noise by exploring an obvious, but powerful, fact: that the schools that children attend matter,” said Kristen Harper, a policy director at Child Trends who worked in the federal education department during the Obama administration. “Their findings are troubling.”

Overall, there’s evidence that students of color are more likely to see their disabilities go unidentified.

The debate about disability rates starts with these key numbers: Nationwide, black and Native American students are a few percentage points more more likely to be classified as having a disability than white students. (Hispanic and Asian students are less likely to have such a designation.)

“We need to address racial and ethnic disparities in special education,” former Secretary of Education John King said in 2016 when the federal rules were rolled out. “This important step forward is about ensuring the right services get to the right students in the right way.”

But others say that overrepresentation is not itself evidence that too many black students are being identified as having a disability. The disproportionate effects of other factors — say, lead poisoning — might affect disability rates in real ways.

“You don’t hear a statistic like ‘minority children are twice as likely to have asthma’ and automatically conclude that pediatricians are racially biased,” Penn State education professor Paul Morgan previously told Chalkbeat.

In a slew of studies, he and a number of other researchers have shown that students of color are generally less likely to be classified as having a disability than similar white students. (This approach does have limitations. Researchers can’t truly measure whether a student “should” receive a disability label, so they use proxies, which are imperfect.)

The new studies in Florida and Wisconsin show that, too.

“Both studies find, as we do, that students of color are on average less likely to be identified as having disabilities than otherwise similar students who are white,” Morgan said.

But students of color, particularly black students, may be over-identified in predominantly white schools.

Where things get more complicated is when you zoom in to what happens to individual students. And that depends on what their schools look like.

In Florida, in predominantly white schools, black students were more likely to be identified as having a disability. In predominantly non-white schools, they were much less likely. There was a similar, albeit less steep, trend for Hispanic students.

Are those differences likely explained by actual differences in disability rates among the students, all born between 1992 and 2002? Probably not, said the researchers.

“What’s amazing about these two studies,” said Harper, “is they try to investigate the influence of those school- and community-level factors, which is incredibly important.”

In predominantly white Wisconsin schools, kids of color are more likely to be labeled as emotionally disturbed or intellectually disabled.

In Wisconsin, it’s white students who seem most affected by being racial “outliers” in their school. If a white student went to a school with few other white students, they were about 10 percentage points more likely to be labeled as having a disability than if they attended a predominantly white school.

But when researcher Rachel Fish of NYU zeroed in on specific disabilities, the picture changed. Looking just at intellectual disabilities and emotional disturbance, Wisconsin starts to resemble Florida. Students of color — especially black students — were much more likely to get such a label if their schools were predominantly white.

Fish’s theory is that this hinges on the different levels of stigma and support attached to various disabilities.

“Rather than being seen as simply low performing or unmotivated, [white students] are more likely than their peers to be sorted into higher-status disabilities,” she wrote. On the other hand, students of color “who are struggling in school are sorted into lower-status disabilities, excluding them from the general education classroom, segregating them with other lower-performing peers.”

In a previous study, Fish found that teachers are more likely to suspect a disability is the root of low academic performance among white students and more likely to suspect a disability is the cause of disruptive behavior for black and Hispanic students.

“There’s always been a view in the civil rights community that schools … end up using special education as a means to create in-school segregation,” said Seth Galanter, an attorney at the National Center for Youth Law. The Wisconsin paper offers some validation for that fear.

Other potential explanations? It could be that teachers are reluctant to identify more than a certain share of their students as having a disability, leading to widespread underidentification at certain schools. It could also be that a student who is a racial minority in a given school stands out more, exacerbating stereotypes and racial bias.

Disability labels seemed to be the most accurate where there was a mix of white students and students of color in a school — in other words, where few if any students stand out solely because of their race. That could be a hidden benefit of integrated schools, and one reason attending an integrated school has been linked to better academic achievement.

“Schools are either more likely to notice actual disabilities or tend to incorrectly apply disabilities to normal behaviors in students who are racially distinct in comparison to the student body as a whole,” the Florida researchers said.