{
  "study_id": "AB-LIBRARY-REGISTER-20261027",
  "version": "1.0",
  "analysis_date": "2026-10-27",
  "question": "What does the current downloadable SNBP 2025-2026 register contain, how consistently can its rows be joined to official municipalities, and what limitations affect reuse?",
  "design": "Descriptive audit of a frozen public administrative-register workbook, not a survey or census of active library services.",
  "protocol_timing": "Written after source acquisition and structural inspection, before the municipal join and substantive aggregate analysis. This is not a preregistered study.",
  "unit": "Non-empty workbook data row. A row is not verified as an independently operating library.",
  "inclusion": "All non-empty rows in Planilha1 of the official July 3, 2026 workbook. Public and community classifications remain separately reported.",
  "exclusion": "The header and entirely blank rows only. Do not remove suspected duplicates silently.",
  "deduplication": "Report exact duplicates and candidate duplicates sharing normalized UF, municipality, library name and address. Retain all source rows unless an exclusion is independently verified and logged.",
  "municipal_join": "Match state plus municipality after Unicode accent folding, uppercasing, whitespace and punctuation normalization. No fuzzy matches in headline counts. Any explicit alias must be separately listed and justified. Brasília/DF is treated as the official municipality-equivalent statistical entry, not multiple municipalities.",
  "population": "Use independently acquired official IBGE 2026 estimates with reference July 1, 2026 if the source is valid. Population-normalized row density is descriptive registry context, not access or service quality.",
  "missingness": "A source value is missing if None or blank after trimming. The initial raw-value postal check converts values to text and trims surrounding whitespace; it ignores Excel number formats and is not a visible-format validity test. Following independent review, inspect the original cell formats: numeric cells using 00000\\-000 are formatted as zero-padded eight-digit CEPs before the display-aware mask check; text values are trimmed. Separately report text values without eight digits after removing non-digits. No test establishes postal-address validity.",
  "outputs": ["national audit measures", "state-level row counts and type breakdown", "matched municipality aggregate table", "unmatched location strings", "source and dictionary version comparison", "machine-readable methods and data"],
  "prohibited_inferences": ["No row means no library", "Register completeness measures reading access", "A registered library is currently open", "Registry density measures service quality", "Type or geography counts establish reading habits", "A version difference proves libraries opened or closed"],
  "verification": "An independent reviewer reopens the source workbook, recomputes key counts, inspects the join and verifies that public wording preserves the unit and limitations."
}