Picture a girl in a village primary school in rural India. She has two teachers. One shares her gender, her caste, and her religion. The other does not. Researchers compared each child’s scores in the subject taught by each teacher, and children did better in the subject taught by the teacher who shared their background. The difference was not small. On average it came to nearly one quarter of a standard deviation, a gain a teacher or a parent would see.
That is ethnic matching, measured in a country where the markers are not race. I start there because last season I spent eleven episodes on what ethnic matching does in the United States, and this season asks what it looks like everywhere else. When the boxes on the form are different, what counts as the match? Who teaches whom, in whose language, and who decided?
What a match has meant
Every country that asks about matching has to decide what to match on, and the answer depends on where its history put the gap. In South Asia the dividing line is often caste. Elsewhere it is religion. In Japan, Canada, Aotearoa New Zealand, and Australia, it is whether the state recognizes a people as Indigenous at all. Across most of the world it is language, which was last week’s subject.
In the United States it has been race and ethnicity. The research has asked what happens for Black students with Black teachers, Latino students with Latino teachers, and Native and Asian students with teachers who share their histories. About eighty percent of the U.S. teaching force is White, and the students are not. That gap has a racial history, and it is the record I covered in Season 3.
So when the gap is different, the match is different, and researchers have to measure something else. Looking at how other countries measure it shows more clearly what the match is.
One child, two teachers
The Indian study looks most like the U.S. pattern, so it is the place to start.
The researchers had data on 5,028 children in rural primary schools in Uttar Pradesh and Bihar, in grades two and four. What made the study unusual was its design. Comparing matched children with unmatched children leaves a doubt: the two groups may have differed from the start. So the researchers compared each child with herself. Same child, two subjects, two teachers. In one subject the teacher shared her gender, her caste, and her religion. The other subject came with a teacher who did not. On average she scored higher in the matched subject.
The researchers explained the result the way the U.S. research does. A teacher who shares a child’s world expects more of that child, and treats her as someone who belongs there. Recognition, expectation, and trust carry the effect. The markers that carry them change from place to place.
I want to be careful with caste. Caste is its own system, with its own history and its own rules, and it should not be read as race by another name. What the Indian study shares with the U.S. research is the finding. When the adult shares the child’s world, the child does better, and the researchers can measure it.
What the research says
So what does the evidence say when it is put together?
In the United States, decades of studies show that when a child has a teacher who shares her racial or ethnic background, several things change. Teachers expect more of her. She is more likely to be referred to gifted programs and less likely to be disciplined, and more likely to finish high school and go on to college. The effects are clearest for Black children with Black teachers, and clearest of all when the match comes early.
The test-score picture is more mixed, and it deserves to be named rather than smoothed over. Some studies find gains, some find none, and a few find small losses for Hispanic students. The authors of the largest of those studies cautioned that the label Hispanic covers very different communities. When the only thing that matches is a label on a form, the match may not be there. The more consistent findings are about how children are treated: what teachers expect of them, who gets referred, and who gets disciplined.
Outside the United States there is far less research. Countries with different histories have different gaps and different markers, and fewer researchers are asking the question. The Indian study is one of a handful. That thin record is one of the reasons this season exists.
Think about your own schooling for a moment. Recall a teacher who understood you. What did you share with that person? And was it something a form would have written down?
Language, the part a district can name
Last week’s subject returns here, because language may be what the label on the form was missing.
A new U.S. study took place in one California district and thirteen elementary schools. It looked at Hispanic students, which is the study’s own category, who shared both their ethnicity and their language with a teacher. Language was measured by whether the teacher held a state bilingual authorization, which requires demonstrated proficiency. Sharing ethnicity alone raised math and reading scores. When the teacher also held a bilingual authorization, the math gains were the largest the researchers measured. It is one district, so it is a starting point and not the last word. Still, it runs in the same direction as the Indian finding.
Consider what that means for a district. Every part of a match depends on who the district recruits and hires, and that includes race and ethnicity, which is why pathways for Teachers of Color matter so much. But language is the part a district can name most directly. It can write language into a job posting, certify it, and build it into how teachers are placed. Of all the parts of a match, language is the easiest to staff for.
One more example, briefly. Japan recognized the Ainu as an Indigenous people by a resolution of its parliament in 2008, and in law in 2019. Recognition was a first step. But Japan’s national statistics ask about nationality, not ethnicity, so a school system has little way to see whether an Ainu child ever has an Ainu teacher. A people the state does not count is very hard to match.
Whoever designs the form
This brings us to the cultural context, because a rule sits underneath this episode and almost nobody states it.
Whoever designs the form decides what a match is. The U.S. form has boxes for race and ethnicity, so U.S. researchers study race and ethnicity. The Indian data included caste and religion, so the researchers could see caste and religion. Japan’s forms ask about nationality, so a match for an Ainu child stays close to invisible. Whatever the form does not write down, the research cannot see, and the district cannot hire for.
Who benefits from a narrow definition? A system that wants to say it has no matching problem can point to its categories and say the numbers show none. If the form never asked, the gap was never written down, and a gap that was never written down does not have to be closed.
On this show I have called the cost the misrecognition tax. Here it is charged before anyone reaches a classroom. It is charged at the desk where someone decides which categories are worth writing down.
The marker is not the match
I have argued for a long time that social knowledge is not like physics. In science, a law lets you predict what will happen under the same conditions every time. People do not work that way. The law of human response starts there: we cannot assume two children will respond the same way to the same conditions, because each one brings her own history, context, and lived experience into the room. That law has a second half. When the person at the front of the classroom can relate to a child in a culturally responsive way, we can better predict how that child will respond, how she will engage with the teacher, and how she will take in the lesson.
A match is a human response. The marker, whether it is race, caste, language, or whatever the form happens to include, is a proxy. It stands in for the chance that the adult will recognize the child, expect something of her, and be trusted by her family. In the United States the proxy is race and ethnicity, because that is where the history is. Rural India looked to caste, religion, and gender. The relationship is the thing. A category is how a country tries to find it from the outside.
None of us is only one thing. Researchers call this intersectionality. A child is a girl, and a member of a caste, and a member of a faith, and a speaker of a language, all at once. The Indian study measured the gain when the teacher shared all three of the child’s markers at the same time. The California study, for its part, found its largest gains when the teacher shared both ethnicity and language. Both looked at overlapping identities and not a single box, and both found real gains.
That is also where misrecognition begins. When a school assumes that a teacher with the same skin tone is automatically a match, it can miss everything else the child brings: her language, her faith, her family’s history, the place she grew up. Skin is not always kin. A shared category on the form is where recognition can start. It still has to happen in the room.
Would it take root?
Each episode this season ends on the same question: would it take root? It comes from Michael Sadler, who argued in 1900 that you cannot take a practice from one country’s schools and expect it to work the same way in another without understanding what holds it up.
Three things come out of this episode. Ethnic matching travels, but the marker does not, so every country has to look where its own history put the gap. Measure how children are treated and whether they belong, and not only how they score, because that is where the evidence is most consistent. And count the groups your form leaves out, because you cannot hire for a gap you have not measured.
What stays local? Caste. The U.S. racial history. Japan’s definition of who counts as Indigenous. The marker is local. The match is not.
If you work with children, here is one thing to try this week. Ask one child what makes a teacher understand them, and write down the answer. Then look at the categories on your form and see whether that answer appears anywhere on it.
Next week the question turns darker. Some systems did not fail to recognize children. They were built to remove them from the adults who could teach them who they were.
