In a Laredo elementary school a few miles from the Rio Grande, the intake paperwork tells one story and the reading scores tell another. Most of the children qualify for free lunch. Many arrive speaking Spanish at home. On the standard demographic model that researchers have leaned on since the 1966 Coleman Report, a school like this is expected to post low proficiency. It does not. Its students read and do math at rates you would sooner predict in a comfortable suburb.
That gap between prediction and result is what allk12's BeatsExpectations metric is built to measure, and when we roll it up to the district level, the districts that beat the odds the most are, with one telling exception, not the wealthy ones. They are places like Laredo ISD.
How BeatsExpectations Works
Within each state, we take every scored school and run a regression of test proficiency on the share of students living in poverty. Poverty is the single strongest demographic predictor of test scores in American education, and the regression draws the line that best describes that relationship across the state. A school's residual is the vertical distance from that line: how many points its actual proficiency sits above or below what its poverty level alone would predict.
A positive residual means the school outperformed the prediction. We then average that residual across all of a district's schools. A district scoring +22 is not scoring 22% on tests. It is landing about 22 percentage points higher than the demographic model expected. The underlying school data comes from the National Center for Education Statistics (NCES) and each state's own assessment results for the 2024-25 school year.
Three things to take note because the metric is easy to over-read. First, BeatsExpectations is a within-state residual. The line is fit separately in Texas, in California, in New York, and everywhere else, so a +20 in one state and a +14 in another were measured against different baselines. Read the ranking as directional, not as a precise cross-state ladder. Second, the New York City entries are individual geographic community school districts, because federal data splits the city that way rather than reporting it as a single system. Third, and most important, a high BeatsExpectations district is not necessarily a high-scoring one. This is a measure of overperformance relative to demographics, not of absolute achievement.
The Large Districts That Beat the Odds the Most
The table below covers large districts only: at least 15,000 students and at least five scored schools, so that a single strong school cannot carry the average. Enrollment and proficiency figures are district-wide; the final column is the average points above predicted.
| District | State | Enrollment | Avg. proficiency | Beats-expectations (points above predicted) |
|---|---|---|---|---|
| Laredo ISD | TX | 20,178 | 50.9% | +22.2 |
| Garden Grove Unified | CA | 36,578 | 54.5% | +20.8 |
| NYC Geographic District #20 (Brooklyn) | NY | 43,965 | 64.2% | +20.2 |
| NYC Geographic District #26 (Queens) | NY | 29,203 | 69.8% | +18.5 |
| NYC Geographic District #25 (Queens) | NY | 33,704 | 63.5% | +15.4 |
| NYC Geographic District #2 (Manhattan) | NY | 54,452 | 70.8% | +14.2 |
| Brownsville ISD | TX | 37,089 | 50.6% | +14.1 |
| Columbia 93 | MO | 17,973 | 62.7% | +14.0 |
| Downey Unified | CA | 22,023 | 46.9% | +13.8 |
| Pharr-San Juan-Alamo ISD | TX | 28,708 | 47.2% | +13.7 |
| Naperville CUSD 203 | IL | 15,672 | 73.5% | +13.4 |
| Everett School District | WA | 19,931 | 60.4% | +12.9 |
| La Joya ISD | TX | 22,864 | 45.2% | +12.9 |
| United ISD | TX | 40,011 | 56.1% | +12.8 |
| Ysleta ISD | TX | 32,245 | 52.5% | +12.3 |
Why the Texas Border Keeps Appearing
Seven of the fifteen districts sit in Texas, and most of those cluster along the border. Laredo ISD, Brownsville ISD, Pharr-San Juan-Alamo ISD, La Joya ISD, United ISD, and Ysleta ISD near El Paso are all high-poverty, overwhelmingly Hispanic districts in the Rio Grande Valley and along the river. Their raw proficiency numbers are not high. Laredo, La Joya, and Pharr-San Juan-Alamo all land in the 45 to 51 percent range. What lifts them to the top of this list is the poverty they are working against. In districts where nearly every family qualifies for free or reduced-price lunch, a 50 percent proficiency rate is far above the line the state model draws.
The pattern is not new to researchers who study Latino education. Work by Roberto Gonzales (2016) and the broader immigrant-optimism literature going back to the 1990s describes how first- and second-generation immigrant families often bring high educational aspirations that show up in effort and attendance even where household income is low. Grace Kao and Marta Tienda argued as early as 1995 that this optimism, more than any single school policy, explains why some immigrant-heavy communities outrun their income profile on academic measures. I would be careful not to romanticize this. These are still districts with real resource constraints, and the metric does not capture graduation pathways or what happens after twelfth grade. But the test-score residual is consistent and large, and it repeats across six independent border districts, which is harder to dismiss than any single school. When the same signal shows up in six separate places, measured against the same statewide line, the case for it being noise gets thin.
Southern California and the Immigrant Cities
The California entries tell a similar story in a different setting. Garden Grove Unified in Orange County, at +20.8, is a large, heavily Vietnamese and Latino district in a working-class stretch of Southern California. Downey Unified, southeast of Los Angeles, is predominantly Latino and also clears +13. Both post proficiency in the moderate range while beating their predictions by wide margins. Neither is the kind of district that lands on a list of California's highest scorers, which is precisely the point of measuring residuals rather than raw results.
The New York City districts extend the theme into a dense urban system. District #20 covers Bensonhurst and Sunset Park in Brooklyn, a heavily immigrant Chinese and Latino area. Districts #26 and #25 sit in northeast Queens, long known for strong schools serving large Asian-American populations. Their raw proficiency is higher than the Texas border districts, in the 63 to 70 percent range, and they still beat expectations by 15 to 20 points because the poverty in their catchment areas would predict lower. Because federal data reports these as separate community school districts rather than as one New York City system, they surface individually here rather than being diluted into a citywide average.
The One District That Does Not Fit
Naperville CUSD 203, in the affluent western suburbs of Chicago, is the exception that clarifies what the metric is and is not doing. Its proficiency, 73.5 percent, is the highest in the table. It is not a high-poverty district. It appears here because even against the high proficiency the state model already predicts for a district that wealthy, Naperville still comes out about 13 points ahead. Its overperformance is real, but it is a different kind of overperformance than Laredo's. One is beating a low bar by an enormous margin; the other is clearing an already-high bar.
That contrast is the reason to read BeatsExpectations as a lens rather than a leaderboard. It surfaces districts doing more with their students than the demographics would forecast, which is a genuinely useful thing to know. It does not tell you which district has the highest scores, the best facilities, or the strongest outcomes ten years later. A parent comparing options should hold the residual alongside the raw proficiency, not in place of it. The same demographically adjusted metric drives our best schools rankings and the studies at allk12 reports, where you can see it applied school by school.
The honest summary is narrow and worth stating plainly. Across a set of large American districts, most of them poor and most of them heavily Hispanic or immigrant, students are scoring well above what a demographic model predicts. The metric cannot tell you why with certainty. But it can tell you where to look, and the where is consistent enough that it deserves the attention.
Sources
allk12 analysis of National Center for Education Statistics (NCES) data and state assessment results, SY 2024-25, using allk12's BeatsExpectations metric.
NCES, Common Core of Data
