Vermont's test scores fell for a decade, on every ruler that spans it. College-going fell sixteen points. Chronic absence rose by half. The graduation rate moved three. Part 1 of 4: the numbers, and the easy explanations that don't survive them.
Part 1 of 4, a series on Vermont's schools statewide. Vermont's test scores have fallen for a decade — on the federal test, on the national district scale, and by the state's own proficiency bars within a single test's era. The share of graduates going on to college or training fell sixteen points. Chronic absence rose by half. The graduation rate moved three points. Three kinds of measures of the same students, and one did not move with the others. This story lays out the numbers, and rules out the easy explanations: the pandemic, the demographics the data can measure, and special education. The three that follow report what Vermont changed in those years — and what its diploma now certifies.
The Gazette's earlier reporting compared Washington Central, and then Vermont, against the country, and found that most of what looks like strength in one district is a lead over a state average that is itself losing ground to the nation. This series is the statewide sequel. It turns on one puzzle: scores, measured several ways, fell; college-going fell with them; attendance worsened; and the graduation rate stood still.
Start with the one test given the same way in every state: the National Assessment of Educational Progress, the federal "Nation's Report Card." A decade ago Vermont sat well above the national average on all four of its checks — reading and math, in grades four and eight. The leads ran from six to twelve points, and a point is roughly a tenth of a grade level. By 2024, three of the four leads were gone. Fourth-grade math sat 2.5 points below the national average, a gap the federal snapshot calls statistically significant. Fourth-grade reading sat about even. Eighth-grade reading sat about even. Only eighth-grade math still held a real lead, at 3.4 points. The absolute falls are large. Vermont's eighth-grade reading average dropped seventeen points between 2013 and 2024. Fourth-grade reading dropped fifteen. That is roughly a grade level and a half.
Stanford's Education Data Archive tells the same story with one number. It places each district — and each state — on the NAEP scale, in grade levels above or below a fixed national norm. Vermont stood about nine-tenths of a grade above that norm in 2013. By 2025 it stood about half a grade below, having crossed the line sometime between 2019 and 2022.
The state's own test agrees, where it can be read. Vermont changed tests three times in eighteen years, so its proficiency rates cannot be compared across the whole period — the second story in this series reports that history. But within the Smarter Balanced era — comparing 2015-16 with 2021-22, the same test at both ends — proficiency fell about twelve points in English and twelve in math.
College-going moved with the scores. Vermont tracks the share of each graduating class enrolled in college or training within 16 months. That share peaked at 62% in the 2020 snapshot. The latest reads 45.6% — a fall of sixteen points. (Each snapshot describes a class that graduated about two years earlier.) The national reference point is 62% — and that figure counts only enrollment by the first fall after graduation, a tighter clock than Vermont's 16-month window, so the comparison can only flatter Vermont.
Attendance moved the same way. Before the pandemic, about one Vermont student in five was chronically absent — missing a tenth of school days or more. The latest count is more than one in four.
Vermont's four-year graduation rate drifted from about 85% in 2017-18 to 82% in 2024-25. On the latest federal count it stood at 83%, and only nine states graduated a smaller share. The country moved the other way: the national rate rose from 80% in 2011-12 to a record 87% in 2021-22. Vermont's students test measurably lower than they did a decade ago, on every measure that can be compared. About the same share of them receive diplomas.
On three of the four national checks, most of Vermont's fall against the country was done by 2019 — before the pandemic closed a single school. The slide began around 2013, from a peak, by several measures, in that very year. The pandemic made everything worse, here and everywhere. It did not start this.
The state's changing population does not explain it either. Stanford's archive pairs each district with a Census measure of the community it serves — income, education, poverty, employment, family structure. On that measure, Vermont's communities grew more affluent relative to the country over these years, which should push scores up. Score every district in the nation against what its community profile predicts, and Vermont's fall gets steeper, not gentler. In 2013, 42 of the 48 Vermont systems the archive can rank beat their expectation. By 2024, 8 of 44 did. The median Vermont system fell from the 81st percentile to the 23rd, after adjustment.
One demographic question survives that check. The Census measure describes whole communities, not the families whose children fill the schools. A town can grow richer while its school-age families grow poorer — call it a K-shaped community: wealthy retirees and childless earners moving in, middle-class families with children moving out, lower-income families with children staying. Vermont's own student data cannot settle this. The state's measure of low-income students reads higher today than a decade ago, but the counting method changed mid-decade, so that line cannot be read straight either. If Vermont's communities went K-shaped, some of the adjusted fall is the yardstick's error, not the schools'. But that reading is no acquittal. It would mean the towns grew richer while the families raising children in them grew poorer — that Vermont's prosperity stopped reaching its children, and the tests were recording it. A different failure, not a smaller one.
Vermont does identify more students for special education than it used to: 19.6% of students had an individualized education program in 2023-24, up from 17.9% four years earlier, and above the national rate of about 15%. Some of that rise may reflect real need, not just changed labeling. Either way, the newly identified students were already enrolled in Vermont schools, already taking the tests, and already counted in every number here. An IEP changes a student's label, not who sits in the classroom. Even if every newly identified student scored forty points below the rest — a wider gap than the state reports — moving two percent of students across it would shift the statewide average by less than one point, against a twelve-point fall on the same test. A tenth of the drop, at most. A 2017 state study did hear anecdotes about families moving to Vermont for its special-education services. But no data bear that out — the school-age population shrank from 2020 to 2024, even through the pandemic moving wave.
The pandemic, special education, and the demographics the data can measure are off the list; the K-shaped question stays open. So does the cause. Scores fell in nearly every state, so some of Vermont's fall is the country's — child poverty, adolescent mental health, staffing shortages. But Vermont fell faster, from higher up — and in the same years, the state rebuilt what a diploma requires, changed its statewide test three times, and moved its schools from an independent board into the governor's cabinet. The next three stories take them up: what a diploma requires, who runs the schools, and what the diploma that held still now certifies.
NAEP samples two subjects in two grades. It says nothing about the rest of what schools do. The demographic adjustment measures communities, not enrolled students; it covers between 30 and 48 Vermont systems in a given year — 48 in 2013 and 44 in 2024, the years quoted here. The college-and-training measure counts Clearinghouse-matched enrollment within 16 months; it misses apprenticeships, military service, and most short certificate programs. The special-education check rules out composition, not rising need: the state's own report allows that some of the increase in identification is real. And ruling out three explanations does not name a cause. That question stays open through this series.
Explore the underlying data yourself: the Comparer · District Outliers.
Method and caveats: NAEP figures are average scale scores for public-school students, NCES NAEP Data Service, with 2024 statistical-significance calls from the NCES state snapshots; one scale point is roughly a tenth of a grade level — a conversion good for magnitude, not precision. Grade-level standings are from the Stanford Education Data Archive 2025.1, state series, against the fixed 2019 U.S. norm. The demographic adjustment regresses, per year, every reporting U.S. district's standing on the archive's Census-based socioeconomic composite (roughly 10,500 in 2024 and 13,600 in 2013, the years quoted) and ranks the residuals; the composite describes communities, not enrolled students. State proficiency: percent proficient or above, grades 3-9, weighted by students tested (AOE files); English 56.5% in 2015-16 to 44.3% in 2021-22, math 45.4% to 33.0%, within the Smarter Balanced era. Graduation: adjusted cohort graduation rate, Vermont's series and NCES Digest table 219.46. College and training: Vermont Annual Snapshot 16-month indicator (61.4% in the 2019 snapshot, 62.0% peak in 2020, 45.6% in 2024); the U.S. 62% is immediate fall enrollment of 2022 completers (NCES). Chronic absence: AOE dashboard counts (18.4% in 2018-19, 27.2% in 2023-24; 2021-22 spans remote and hybrid attendance). Special education: the Agency's Act 73 special-education report (2025) — IEP share 17.9% (2019-20) to 19.6% (2023-24) against a national rate near 15%; the identified count rose while enrollment fell; the report attributes the rise to identification practice while allowing that some reflects rising need. Relocation anecdotes: the state's 2017 special-education funding study (unquantified practitioner interviews); school-age population per the Joint Fiscal Office (2024). That newly identified students remain in the tested averages follows from federal participation rules — at least 95% of students tested, a 1.0% alternate-assessment cap — an arithmetic observation, not a state finding; so is the magnitude bound (a 1.7-point share shift times a forty-point proficiency gap is under 0.7 points, against the twelve-point Smarter Balanced-era fall). The K-shaped-community question is likewise open arithmetic, not a finding: the enrolled-student low-income series cannot be read across the decade because the measure changed (free-and-reduced-lunch counts broke with universal school meals; the Act 127 direct-certification count begins later), and the Census composite describes whole communities.
Sources: NCES NAEP Data Service and 2024 State Snapshots; Stanford Education Data Archive 2025.1 (state series and ACS covariates); NCES Digest 219.46; Vermont AOE assessment files, graduation series, Annual Snapshot, attendance dashboard; Vermont AOE Act 73 special-education report (2025); Vermont special-education funding study (2017); Vermont JFO demographics (2024); 20 U.S.C. 6311(c)(4)(E); 34 CFR 200.6(c)(2).
The Middlesex Gazette is a one-man paper, kept by a neighbor for his neighbors. A town is best governed when its doings are known in every kitchen — so here they are, set down plain and free of charge. Back to the front page →
The Gazette comes free by email — a note when there's town news worth knowing. Subscribe by email →