Universal Pre-K kids are now in grade 5. Did the investment hold?
The question
Has New York City's Universal Pre-K, rolled out in 2014, lifted students' test scores a decade later?
We can't directly answer whether New York City's Universal Pre-K (rolled out starting in 2014) lifted later test outcomes. Citywide fifth-grade math went from 45% in 2018-19 to 54% in 2024-25, but most of that gain comes from the state changing how it scores the test in 2024-25. The cleanest analysis would compare neighborhoods where Universal Pre-K rolled out faster with neighborhoods where it rolled out slower — possible with public data we haven't yet pulled in. Published research (MDRC's 2022 evaluation, Christina Weiland's Boston work) finds modest gains in kindergarten readiness that mostly fade by third grade.
Data analysis
Investigation run 2026-05-25.
TL;DR
We can't isolate the Universal Pre-K effect from our current data. Fifth-grade outcomes in 2024-25 for the 2014-15 Pre-K cohort are entangled with the state's test rescaling, COVID disruptions, and curricular changes. Doing this story properly requires either student-level identifiers (we don't have them) or a neighborhood-level analysis using the speed of Pre-K rollout as a comparison variable (possible with public data we'd need to ingest).
What we can say from our data
Citywide grade-5 math proficient: 44.6% (2018-19) → 54.3% (2024-25). That's +9.7 points. The 2018-19 baseline cohort was 4-year-olds in 2017-18 (had UPK if their family enrolled). The 2024-25 cohort was 4-year-olds in 2023-24 (definitely had UPK access citywide).
So the change spans a population that had partial UPK to a population that had full UPK. If UPK has a long-run effect, we'd expect a positive trend. We see one. But:
- About 7-9 of those 9.7 points is the test recalibration in 2024-25 (citywide grade-3 math jumped 12 points in one year, larger than any UPK literature would predict). - Subtract test recalibration and you're left with ~2 points over 6 years — within noise for this kind of aggregate comparison. - Best we can say: the data is consistent with a small positive UPK effect, but also consistent with no UPK effect.
Why a real answer requires different data
Christina Weiland's Boston UPK work uses regression discontinuity (kids born just before/after the eligibility cutoff). NYC's UPK was universal-then-not (no clean cutoff), so RD doesn't work.
Better approach: instrumental variable using UPK rollout speed. NYC didn't roll UPK out evenly — some neighborhoods had access faster than others. Cross-reference UPK access rate per ZIP code in 2014-17 with grade-5 outcomes in 2024-25.
To do this we'd need: 1. NYC DOE UPK enrollment per ZIP code per year, 2014-2024 (not in our DB but in NYC Open Data). 2. Census tract child population 2014-2024 (for normalization). 3. A way to track which ZIPs had UPK access first.
Best available indirect evidence
MDRC's 2022 UPK evaluation reported modest gains for participants on kindergarten readiness (~0.15 SD), with effects fading by 3rd grade. That matches the broader Pre-K literature: short-run gains, persistence is hard.
If the long-run NYC effect is similar to the literature (~0 to +0.1 SD by 5th grade), it would translate to roughly 1-2 percentage points on the grade-5 math proficient rate. That's within the noise of our aggregate-level data and impossible to isolate from concurrent changes.
The honest answer to "did UPK work?" is: yes for kindergarten readiness, mostly faded by 5th grade based on national literature, and we can't independently verify the NYC-specific effect from our current data.
What would change the answer
Ingest UPK enrollment by ZIP × year (NYC Open Data has 'Universal Pre-K Locations' dataset).
Build the rollout-speed instrumental variable.
Compare ZIPs with high vs low UPK access in 2014-16 on grade-5 outcomes in 2023-25.
If the high-UPK-access ZIPs show 1-3 points more outcome gain than low-access (after controlling for income), that's the UPK effect.
This is feasible — a 1-2 week analysis project once the data is loaded.
Caveats
Aggregate-level analysis (our current dataset) can't disentangle UPK from concurrent changes. The honest framing: 'The promise of UPK was outcomes through 5th grade. We can't yet say whether NYC's UPK delivered, but we have a path to find out.'
Methodology & replication recipe — the data sources, queries, and steps behind this analysis.
More on specific schools
- Six NYC schools where 100% of teachers say students bully each other
- The Brooklyn New School lost 52 points of math proficiency in six years
- Bronx Aerospace High School cut chronic absenteeism by 34 points. How?
- The Anderson School scores at the ceiling — and proves nothing
- Two MS within a half-mile: one screens, one zones. Same neighborhood, different futures.