← Data quality

Validation reports

Machine-generated, per-metric reports from the validation harness. Each row links to the full narrative write-up on GitHub.

Base case (ii) — served values vs the ingested source file

An independent re-read of the exact file each loader ingested, compared cell-by-cell against what the site serves. This is the binding correctness gate: PASS means zero value mismatches. "Cells checked" excludes cells the publisher suppressed for student privacy.

MetricCells checkedMismatchesAgreementStatusReport
chronic_absenteeism_rate
tolerance ±1pp
10,6350100%PASSfull report
ela_all_proficiency
tolerance ±1pp
6,5650100%PASSfull report
ela_grade3_proficiency
tolerance ±1pp
4,6460100%PASSfull report
ela_grade7_proficiency
tolerance ±1pp
2,7570100%PASSfull report
graduation_rate_4yr
tolerance ±1pp
3,7360100%PASSfull report
math_all_proficiency
tolerance ±1pp
6,4490100%PASSfull report
math_grade3_proficiency
tolerance ±1pp
4,6430100%PASSfull report
math_grade7_proficiency
tolerance ±1pp
2,7480100%PASSfull report
survey_student_2021-22
tolerance ±1pp
73,2410100%PASSfull report
survey_student_2022-23
tolerance ±1pp
81,8490100%PASSfull report
survey_student_2023-24
tolerance ±1pp
76,5610100%PASSfull report
survey_teacher_2021-22
tolerance ±1pp
227,2530100%PASSfull report
survey_teacher_2022-23
tolerance ±1pp
185,0840100%PASSfull report
survey_teacher_2023-24
tolerance ±1pp
182,3660100%PASSfull report

Spot check (iii) — vs NYC's School Quality Snapshot

A sampled comparison against NYC's public per-school report. The Snapshot reports a different population cut of the same underlying data — the documented accountability cohort for the test metrics; for chronic absenteeism, the grade-band rate (K–8 / 9–12, finding A.6) — so disagreement here is recorded as CONTEXT, not failure. The chronic-absenteeism sample below is superseded by an exhaustive school-by-school census on the band basis (snapshot/CENSUS.md); the remaining tails are tracked on the findings ledger. We do not adjust served values toward the Snapshot.

MetricCells checkedMismatchesAgreementStatusReport
chronic_absenteeism_rate
tolerance ±1.5pp
2127166.5%CONTEXTfull report
ela_all_proficiency
tolerance ±1.5pp
2957973.2%CONTEXTfull report
math_all_proficiency
tolerance ±1.5pp
29010962.4%CONTEXTfull report

Cross-publisher reconciliation (check iii-b, NYSED) has its own figure-rich page and report directory. Methodology for all checks: verify/METHODOLOGY.md.