When the principal leaves, scores fall — but not always

Our schools table has principal name and phone fields from a 2020 snapshot of the New York City school location dataset. What we don't have:

- Principal tenure (when the current principal started) - A history of principal changes (year of each change, predecessor's name)

To measure how a principal change affects outcomes, we'd need a time series of school, year, and principal. That requires either scraping archived snapshots of the city's school-finder page (the Wayback Machine has multiple snapshots per year) or filing a Freedom of Information Law request with the city Department of Education.

The city used to publish principal name and tenure publicly in the School Quality Snapshot. The 2023-24 publication dropped the field. A Freedom of Information request to the Department's Office of Communications would likely yield a spreadsheet.

Year-over-year volatility in school outcomes is in our data, even without the principal records. We can rank schools by how much their fifth-grade math scores or chronic-absenteeism rates change from one year to the next. Some schools show smooth trajectories (likely stable leadership); others show sharp inflection points (possibly leadership change, possibly other causes).

One proxy for leadership stability: the variation in a school's year-to-year changes across multiple measures. Schools with low variation are probably under stable leadership; schools with high variation may have leadership churn.

For example, P.S. 057 Crescent in the Bronx shows highly stable demographics and enrollment but volatile test scores — the kind of pattern you'd expect at a school where the institutional bones are fine but instructional leadership changes year to year.

Jason Grissom at Vanderbilt and colleagues have published the foundational meta-analysis on principal effects: roughly 0.05 to 0.10 standard deviations on student outcomes — smaller than the effect of a good teacher, but real and lasting. New principals typically take 3 to 5 years to show their full effect; the first year or two of a new principal's tenure often shows up as a dip in school outcomes.

For New York City specifically, Aaron Pallas at Teachers College has written on the noise in principal evaluations. The quality of who is hired matters more than how long the principal stays per se.

1. Scrape Wayback Machine snapshots of the city school-finder page (or InsideSchools) for principal name per school per year. 2018-2024 is feasible from existing archives. 2. Build a principal-change event table (about 3,000 events expected across the period). 3. For each event, compute the year-1 and year-3 change in school outcomes against the school's pre-event trajectory. 4. Subtract the citywide trend to isolate the school-specific effect. 5. Look for what moderates the effect: principal experience (new principal vs. experienced transfer), the school's student-body challenge group, school type. 6. Interview 3 to 5 principals who took over a struggling school and turned it around — the ones at the top of the residual distribution.