Why does middle school cause the academic cliff?
The question
Why do test scores drop sharply between elementary and middle school in New York City, and does the structure of the school students attend matter?
Citywide math proficiency falls 18 percentage points from third grade (62%) to eighth grade (45%), with the steepest single drop between fifth and sixth grade (54% to 44%). The middle-school cliff is real. K-8 schools' seventh-graders score about 4 percentage points higher in math than their counterparts at standalone middle schools (55.5% vs. 51.7%), which is some support for the 'transitions are bad' hypothesis — though families who choose K-8 schools may also be more engaged, which would produce the same gap through selection.
Data analysis
Investigation run 2026-05-31.
TL;DR
The middle-school cliff is real in this data. Citywide math proficiency drops from 62% in third grade to 53% in seventh and 45% in eighth — about 17 percentage points lost across that span. K-8 schools handle the cliff better than standalone middle schools: their seventh-graders average 55.5% proficient in math against 51.7% at standalone middle schools. That's some support for the idea that grade-to-grade transitions hurt performance, though selection effects make the gap hard to pin down.
Citywide proficiency by grade (2024-25)
Math drops 18 percentage points from third grade to eighth grade. Reading is steadier (a 5-percentage-point spread). The middle grades — 6 and 8 — score lower than the grades on either side, so the cliff isn't a smooth slope but a series of dips. Sixth-grade math (44%) is well below fifth-grade (54%) and seventh-grade (53%) — a transition-year dip rather than a sustained downhill.
K-8 schools versus standalone middle schools
K-8 schools' seventh-graders score about 4 percentage points higher on math than seventh-graders at standalone middle schools. That's some support for the 'continuity of relationships and culture' hypothesis. The obvious alternative explanation: families who choose K-8 schools may be more engaged on average, which would produce the same gap through selection rather than through the school structure.
| School type | Schools | Average seventh-grade math proficient |
|---|---|---|
| K-8 schools | 133 | 55.5% |
| Standalone middle schools | 238 | 51.7% |
What's happening in sixth-grade math
Sixth-grade math at 44% proficient is the lowest single grade in the data. Math changes character at sixth grade — algebra preparation, more abstraction. The drop coincides with the cohort shuffle at standalone middle schools: students arrive in a new building, with new teachers and new peers, and face a harder curriculum at the same time. All four changes hit at once, and the size of the dip matches.
Caveats
A same-cohort comparison would be the ideal test — track the 2018-19 third-grade cohort through to eighth grade in 2023-24 — but we have school-level rather than student-level data. The cross-grade comparison above mixes cohorts and curricula together.
What's next
Worth a piece on the sixth-grade dip specifically. New York City's K-8 expansion in the early 2000s was partly motivated by this cliff. Re-examining whether the expansion worked is overdue.
Methodology & replication recipe — the data sources, queries, and steps behind this analysis.