Schools that brand themselves 'STEM' don't outperform other schools on math
The question
Does a STEM label on a New York City elementary school correspond to better math outcomes, or is it mostly branding?
Sixteen New York City elementary schools label themselves STEM-focused. Their average fifth-grade math proficiency is 56.7%, against 54.2% at the 765 schools without the label — a 2.5-percentage-point gap, well within typical school-to-school variation (the school-level spread is about 20 points). The label doesn't track to a measurable difference in math outcomes at the elementary level. That doesn't mean STEM programs are useless — it means the basic math proficiency rate isn't where the difference shows up.
Data analysis
Investigation run 2026-05-31.
TL;DR
Sixteen New York City elementary schools label themselves STEM-focused. Their average fifth-grade math proficiency is 56.7%, against 54.2% at the 765 schools without the label — a 2.5-percentage-point gap, well within the normal school-to-school spread. At the elementary level, the STEM label doesn't track to measurable differences in math outcomes.
STEM-labeled vs. other elementary schools
| Group | Schools | Average grade-5 math proficient |
|---|---|---|
| STEM-labeled elementary schools | 16 | 56.7% |
| Other elementary schools | 765 | 54.2% |
| Difference | — | +2.5 pts |
What the gap means
A 2.5-percentage-point difference is roughly one-eighth of the normal school-to-school spread. Statistically you'd only call it meaningful at very large sample sizes. Practically, for a parent choosing a school, it's not a difference you'd notice. A STEM label at the elementary level is signal, not measurable substance.
What this analysis doesn't measure
The fifth-grade math proficiency rate may not be the right outcome for judging STEM programs. A STEM-focused elementary school might produce students who choose STEM electives in high school, join robotics clubs, or graduate with stronger computer-science skills. None of those outcomes are in our database. We're testing the weakest version of the hypothesis (basic math test scores) and finding nothing; the stronger version (long-run STEM career persistence) we can't test from this data.
Where STEM branding probably does track to outcomes
At specialized high schools — NYC iSchool, Brooklyn STEAM Center, Stuyvesant — STEM focus does correspond to strong outcomes. But those schools enroll self-selected students who chose them for the focus. The lift comes from the selection, not the branding alone. Elementary-school STEM labels try to capture the same effect without the selection mechanism, and the data suggests they don't move the proficiency rate.
What's next
Story angle: STEM branding at the elementary level is marketing, not measurable curriculum impact. Pair with a profile of NYC iSchool to contrast.
Methodology & replication recipe — the data sources, queries, and steps behind this analysis.