AFGHANISTAN EDUCATION ATLAS — DATA NOTES
Prepared 17 September 2026. Independent research dashboard, not an official census.

FILES
province-indicators.csv / indicators.json: 34 provinces plus the published national estimate.
Percentages: primary (adjusted net attendance, ages 7–12), lower (ages 13–15), upper (ages 16–18). Ages refer to the beginning of the school year. Attendance at the relevant level or higher counts. internet/mobile/computer refer to households, not students. All observations from MICS 2022–23.
Printed source pp. 7, 48; national inequality comparisons on pp. 8 and 47.
School attendance figures describe survey reference periods; they are not evidence of present-day permission for girls to attend secondary school.

schools.json / schools.csv: compact school-location records with source IDs and unchanged coordinates rounded to six decimals. Text fields use first nonblank value from alternative source columns. Original source IDs preserved. No missing locations or enrolment figures invented. Observation year is unknown. No claim that a record is a currently open school. Schools are not students; no individual learner records are included.
schools-source.geojson: original assembled source response, 9,952 records from five ordered pages. 359 coordinate groups contain multiple records; retained rather than guessing duplicates. These are record counts, not guaranteed unique schools. School-ID duplicates and location accuracy need further validation.
quality-audit.json: validation results and province record counts. A province with no source records has unknown coverage, not zero schools.
provinces.geojson: geoBoundaries simplified ADM1 geography. Display spelling harmonised through documented aliases, including source “Ghanzi” to Ghazni. Geometry is not altered. No spatial join is used to invent school province labels.
sources.json: source register, URLs, periods, uses and limitations.
headlines.json: dated national estimates/claims. Do not add overlapping populations or subtract unlike measures.
school-source-metadata.json / boundary-source-metadata.json: captured publisher metadata.

METHOD
MICS numeric tables manually transcribed from the publisher's parsed report. Range, row-count and province joins checked. No modelled, imputed or synthetic values. This is not independent survey validation. National estimates are taken from the report, not calculated as unweighted averages of provinces. No standard errors are included; do not interpret small differences as statistically significant. No causal claim connects internet availability to attendance.

RIGHTS AND ATTRIBUTION
Original reports remain linked at publishers. Numerical facts are attributed. UNDP layer provides no explicit licence text; do not interpret public access as an open licence for downstream redistribution. Consult publisher terms. geoBoundaries metadata labels boundaries Public Domain. Interface code is separate from third-party data rights. Leaflet BSD-2-Clause and Leaflet.markercluster MIT notices supplied in vendor files.

SCOPE GAP
This edition maps the 9,952 available records across 26 provinces. It does not claim to map all 18,337 reported schools. Completing a national inventory requires a newer authoritative school register, coordinates and validation. No percentage coverage is calculated against a total from a different source/date.

STREET BASEMAP
The dashboard uses live OpenStreetMap Standard tiles (https://tile.openstreetmap.org/{z}/{x}/{y}.png) for roads, place names and mapped buildings. Copyright OpenStreetMap contributors; data licensed under ODbL: https://www.openstreetmap.org/copyright . Tile service policy: https://operations.osmfoundation.org/policies/tiles/ . Browser caching is used; no offline tiles, bulk downloads or background prefetching are offered. Tiles are not included in this bundle. Map detail depends on community coverage and internet availability. School coordinates remain the separate UNDP source positions; building-level accuracy is not verified. Automatic mode shows streets in the school layer and from zoom 10 in other layers; maximum zoom is 19.

ENGLISH SEARCH ALIASES
school-search-aliases.json and school-search-aliases.csv contain derived search aids keyed to original school record IDs. Common words use conventional Roman spellings; other tokens receive approximate letter-based romanization, which cannot reliably recover unwritten vowels. Common school-type terms also receive English search variants (for example, Neswan / Girls school). These are not publisher-supplied or official English names and do not verify the school's identity, status or coordinates. Original records, names and coordinates remain unchanged. The 715 unnamed records receive no aliases.
Shams London Academy is a search variant for the source names at IDs 1712 and 6661; the latter is spelled Shamsh in the source. This does not add or verify a Kabul branch. Search ignores punctuation, diacritics, Arabic/Persian variants of kaf and yeh, and common Latin vowel/repeated-letter differences for consonant keys of at least three letters. Approximate matches may be ambiguous; compare the original name and province. Build script: scripts/build_search_aliases.py. Broader manually verified English spellings can be added later.
