# NYSED cross-publisher reconciliation — chronic absenteeism

The first **(iii-b) different-publisher check** for NYC: NYSED's Report Card
Database (`ACC EM/HS Chronic Absenteeism`, school-level, BEDS-keyed) vs the
served values ingested from NYC DOE's End-of-Year file, joined per school via
the deterministic BEDS→DBN rule (district + D75; charters need a crosswalk and
were excluded here).

Contents:
- `FORMAT_NOTES.md` — NYSED source scouting: files, tables, columns, years,
  coverage, BEDS structure, suppression (every claim sourced).
- `RECONCILIATION.md` — the comparison itself: all comparable years (2018-19,
  2020-21 → 2024-25) × grades 1–8 (EM) / 9–12 (HS); agreement stats,
  enrollment ratios, worst cases, hand-verified examples. §9 = the 2026-06-10
  expansion (2020-21/2021-22/2022-23 panels, the COVID-era HS tail).
- `NARROWING.md` — document review of the 2024-25 EM divergence: what is
  ruled out, what remains, next step.
- `stats.json` — machine-readable headline numbers.

Status: produced by one-off reconciliation scripts ahead of the NYSED ingest
(see docs/design/06); the verify/ harness gains a NYSED source connector in
that PR, after which these regenerate like other qa_reports. Raw extracts live
under `data/reference/sources/nysed_src/<date>/` (gitignored; sha256 of each
SRC zip recorded in the reports).

Headline: ten of twelve (year × grade-band) panels — all comparable years,
2018-19 and 2020-21 → 2024-25 — reconcile at the median (median per-school
diff ≤0.4pp). The two failures: **2024-25 grades 1–8** (DOE falling vs NYSED
rising on the same schools, numerator-side, weighted 29.0% vs 30.8%, median
−1.84pp; no precursor — EM medians stay within ±0.4pp in every earlier panel;
no documented definition change on either side explains it; cause not
established; inquiry to NYSED drafted; treat 2024-25 EM chronic absenteeism as
publisher-dependent until resolved) and **2022-23 grades 9–12** (median
−0.60pp, only 47% within ±1pp, r 0.779). The 2022-23 HS failure is the visible
edge of a broader **COVID-era HS tail**: 62/112/129 high schools sit >5pp
below NYSED in 2020-21/2021-22/2022-23 on largely byte-identical denominators
— publisher-level, hidden by median-based summaries; cause not established
(RECONCILIATION.md §9; ledger A.20).