NYC's English Language Learners are 32 points behind on math — within the same schools

At the 310 New York City elementary schools where both groups have enough students to report scores, the never-needed-services group scores more than double the currently-in-services group on fifth-grade math. About 50 schools — one in six — have within-school gaps above 50 percentage points.

We only measure the gap at schools where both groups of students have enough kids to report scores — that's roughly 270 to 360 schools per year. Across years, the gap has been mostly stable, with a small rise from 29 points in 2017-18 to 32 points in 2024-25. The most recent year's absolute scores for both groups are higher than they would have been before the New York State test changes (covered in a separate story on test recalibration), but the gap inside the same school is unaffected by that — both groups are measured on the same test on the same scale in the same year.

The reading test is in English; students learning English are still building academic vocabulary in their second language, so they'd be expected to score lower on a reading test. Math is more language-neutral, but the math test is still given in English, with limited translation help. If language alone explained the gap, the reading version would be much bigger than the math version. It is — but math still shows a 32-point within-school gap, which language alone can't explain.

The largest within-school gaps are at schools where the never-needed-services group scored 70-92% proficient while the currently-in-services group scored 0-19%. All 15 of these schools are in Brooklyn, Queens, or Manhattan, and most are in school districts that combine heavy recent-immigrant enrollment with long-established middle-class neighborhoods: District 2 (Manhattan downtown), District 20 (Bay Ridge), District 21 (Coney Island/Gravesend), District 26 (Bayside), and especially District 30 (Astoria/Long Island City). A caveat on the smaller schools: when a school's group of English-language-learner fifth-graders is only three or four students, the percentage can shift dramatically if a single student scores well. So the very precise-looking 0% or 7% should be read as 'very few students passed,' not as those exact numbers.

Queens districts dominate the top of this list. District 30, which covers Astoria and Long Island City, averages a 45-point within-school gap across its 22 elementary schools. District 26 in Bayside averages 43 points across 10 schools. These are districts with substantial recent-immigrant enrollment from many countries — Greek, Bangladeshi, Latin American, Chinese, and others — often in schools that also serve a long-established middle-class or wealthy population that doesn't need language services. The wide within-school spread reflects that combination.

Schools that screen applicants by grades, attendance, and sometimes English-language ability have the narrowest within-school gaps. That fits: those schools are also screening out students who would be in language services. Zoned neighborhood schools — by far the most common type in New York City — have the full 32-point gap. The 'composite' category (a mix of selection methods, mostly older middle schools) shows the largest gap, but the sample is only six schools.

The data can't separate the possible explanations; school-level reporting could. Four candidates:

• The language of the test itself. Math word problems require reading English; multi-step questions stack the reading load. The 18-point difference between the math gap and the reading gap is consistent with language explaining some — but not all — of the math gap.

• How English learners are scheduled. Students learning English often receive separate language instruction (the program is called English as a New Language, or ENL) during portions of the school day — sometimes during the same period when other students are getting core math instruction. New York City has been contracting its older bilingual programs and expanding the pull-out ENL model since around 2010, which means scheduling tradeoffs vary by school and district.

• Who counts as a current English learner. New York City's official label groups together students in their first year of language services with students in their fifth or sixth year. Their starting points and abilities are very different. Breaking down the data by years-in-services would tell us whether the gap closes as students gain more years of English exposure.

• Out-of-school differences. Many never-needed-services students come from families with extensive tutoring, test prep, or other enrichment. The students currently in language services less often do.

A concrete reporting move: visit P.S. 95 in Gravesend (largest gap in the city) and a school in District 5 (Harlem) or District 15 (Park Slope) where the gap is small, and compare how each scheduling approach plays out day-to-day.