Validation — outcome metrics
(ii) Base case
| metric | years | common | matches | mismatches | suppressed | status |
|---|---|---|---|---|---|---|
pct_ell | 2018-19, 2021-22, 2022-23, 2023-24, 2024-25, 2025-26 | 3075 | 576 | 0 | 2499 | PASS |
pct_ell_lower_bound | 2018-19, 2021-22, 2022-23, 2023-24, 2024-25, 2025-26 | 3075 | 3075 | 0 | 0 | PASS |
pct_ell_upper_bound | 2018-19, 2021-22, 2022-23, 2023-24, 2024-25, 2025-26 | 3075 | 3075 | 0 | 0 | PASS |
ela_achievement | 2018-19, 2021-22, 2022-23, 2023-24, 2024-25 | 2273 | 2273 | 0 | 0 | PASS |
math_achievement | 2024-25 | 462 | 462 | 0 | 0 | PASS |
pct_econ_disadv | 2024-25 | 469 | 469 | 0 | 0 | PASS |
chronic_absenteeism_rate | 2017-18, 2019-20, 2022-23, 2024-25 | 2048 | 2037 | 0 | 11 | PASS |
absent_21plus_rate | 2016-17, 2024-25 | 1013 | 1005 | 0 | 8 | PASS |
graduation_rate_4yr | 2023-24, 2024-25 | 283 | 240 | 0 | 43 | PASS |
fast_ela_g3_proficiency | 2023-24 | 299 | 292 | 0 | 7 | PASS |
fast_math_g8_proficiency | 2024-25 | 198 | 177 | 0 | 21 | PASS |
eoc_alg1_proficiency | 2024-25 | 289 | 273 | 0 | 16 | PASS |
science_g5_proficiency | 2022-23 | 294 | 289 | 0 | 5 | PASS |
science_g8_proficiency | 2022-23 | 188 | 174 | 0 | 14 | PASS |
survey_students_safety | 2019-20, 2021-22, 2024-25 | 1440 | 1238 | 0 | 202 | PASS |
survey_parents_teaching | 2018-19, 2021-22, 2023-24 | 1345 | 1113 | 0 | 232 | PASS |
survey_staff_leadership | 2020-21, 2021-22, 2022-23 | 1393 | 1249 | 0 | 144 | PASS |
survey_staff_safety | 2024-25 | 459 | 410 | 0 | 49 | PASS |
survey_staff_teaching | 2021-22, 2024-25 | 924 | 830 | 0 | 94 | PASS |
school_grade_points_pct | 2022-23, 2024-25 | 937 | 937 | 0 | 0 | PASS |
science_achievement | 2022-23 | 452 | 452 | 0 | 0 | PASS |
learning_gains_math_l25 | 2023-24 | 446 | 446 | 0 | 0 | PASS |
(iii-b) Cross-publisher — FLDOE absent-10%+ vs EDFacts DG814, 2022-23
Verdict: PASS — no-PK schools within ±1: 253/264 (95.8%); PK schools' gap = 0.661 × PK membership (r = 0.86). Detail: __cross_publisher_absenteeism.md.
(iii-b) Cross-publisher — FLDOE race shares vs NCES CCD
Verdict: PASS — pct_black: 90.3% within 2 pp (median Δ +0.00); pct_hispanic: 89.7% within 2 pp (median Δ +0.00); pct_white: 88.1% within 2 pp (median Δ +0.00). Detail: __cross_publisher_race.md.
Residual-z checks
Verdict: PASS — independent N, eligibility and refit: max |Δz| 4.1e-11 over 46 fits; served = committed: 19457 cells, 0 CSV-only, 0 served-only. Detail: __residual_checks.md.
(iv) Served-value invariants
Verdict: REVIEW — range violations 0, identity violations 0, stale rows 0, shift alarms 146 (16 unexplained). Detail: __served_invariants.md.