It belongs to Government Quality, Democracy, and Justice, the part of the welfare catalog concerned with public accountability, institutional capability, participation, and administrative performance. The exact stored unit is score from 0 to 1, the geographic presentation is global, the data publisher is World Bank, and the immutable numerical identity is DATA360 VDEM_CORE The preliminary catalog candidate 82 is reconciled to the exact official series “Clean elections index” in score from 0 to 1. This is a deliberate canonical correction: the database title names what the values actually measure, while the input wording remains in the reconciliation and audit trail. No hidden proxy label, fallback series, or silently substituted denominator remains. For catalog item 82, the provider definition is: Question: To what extent are elections free and fair? Clarification: Free and fair connotes an absence of registration fraud, systematic irregularities, government intimidation of the opposition, vote buying, and election violence. Scale: Interval, from low to high (0-1). That definition controls the denominator, population, reporting instrument, and statistical meaning. This Indicator records an observed outcome; it does not assert why that outcome changed. Welfare assessment for item 82 uses higherisbetter. Higher observations are treated as better welfare relative to the dated comparative reference. The direction applies to the stored official measure after the disclosed transformation, if any, rather than being inferred from an emotionally positive or negative title. No mathematical transformation is applied before persistence. Keeping title, series identity, unit, scope, transformation, and interpretation together prevents the record from mixing incompatible measurement concepts. The numerical observations for catalog item 82 were retrieved from V-Dem Core through https://data360api.worldbank.org/data360/data?DATABASEID=VDEMCORE&INDICATOR=VDEMCOREV2XELFREFAIR&timePeriodFrom=2018&timePeriodTo=2025 (opens in a new tab). The provider landing page is https://data360.worldbank.org/en/indicator/VDEMCOREV2XELFREFAIR (opens in a new tab). The research package records the access date, exact dataset and series identifiers, dimension key AGE=T|COMPBREAKDOWN1=Z|COMPBREAKDOWN2=Z|COMPBREAKDOWN3=Z|FREQ=A|SEX=T|URBANISATION=T, unit, provider metadata, transformation, and yearly aggregation label. The bounded Source and Evidence records can therefore reproduce which numbers were used without treating a generic homepage as row-level proof. Measurement provenance for item 82 follows this provider method: The World Bank Data360 API supplies country observations for the exact database, series, and dimension tuple. Factrail records a disclosed unweighted annual mean and preserves the dataset version metadata. 2018 combines 174 reporting economies; 2019 combines 174 reporting economies; 2020 combines 174 reporting economies; 2021 combines 174 reporting economies; 2022 combines 174 reporting economies; 2023 combines 174 reporting economies; 2024 combines 174 reporting economies. The output keeps full numerical precision through NormalLine, deviation, and interpretation calculations. Display clients may round for readability, but the persisted value is not squeezed into a zero-to-one score and is not rounded before calculation. The item 82 NormalLine starts on 2018-01-01 at 0.866 score from 0 to 1. 75th percentile of provider-published reporting-economy observations in the selected series and dimensions. The calculation keeps the source unit, dimensions, and transformation fixed and is a comparative reference rather than an invented statutory target. For zeroisnorm, the value is exactly zero and is never relaxed because non-zero observations are common. For every item 82 point, normallinevalue copies that dated line and deviationfromnormal is stored in raw-unit terms as observation minus NormalLine. Arithmetic sign and welfare sign are separate. A positive raw deviation is adverse for lowerisbetter, favourable for higherisbetter, and adverse above zero for zeroisnorm. The custom governance score retains its provider scale and uses the public methodology explanation rather than title sentiment. The complete retained sequence is 2018: 0.5567528736 score from 0 to 1; 2019: 0.5544942529 score from 0 to 1; 2020: 0.549137931 score from 0 to 1; 2021: 0.5392298851 score from 0 to 1; 2022: 0.5304367816 score from 0 to 1; 2023: 0.5351091954 score from 0 to 1; 2024: 0.5265344828 score from 0 to 1. The first observation is 0.5567528736 score from 0 to 1 in 2018 and the latest is 0.5265344828 score from 0 to 1 in 2024. The absolute change is -0.0302183908 score from 0 to 1, equivalent to -5.42761290204% of the absolute starting level. The minimum observed value is 0.5265344828 score from 0 to 1, and the maximum is 0.5567528736 score from 0 to 1. For catalog item 82, the latest observation differs from the NormalLine by -0.3394655172 score from 0 to 1 and is classified as negative under higherisbetter. That classification is a deterministic comparison, not a causal explanation or editorial verdict. A movement may reflect real change, altered reporting coverage, revised estimates, a changed survey frame, or several of those at once. There are 7 retained item 82 observations in the selected 2018-2025 research window. The retained observations satisfy the minimum temporal rule without interpolation, extrapolation, or forward filling. Missing years are absent observations, not zeroes. The sequence is intentionally irregular when the official source is irregular, because a smooth-looking line created by invented values would be methodologically false. The displayed global item 82 sequence has one consistent numerical identity: DATA360 VDEM_CORE No second source series is spliced into a gap, no latest value is copied into a missing year, and no country-only observation is relabelled as a World value. If the source later adds or revises an observation, the update must arrive as a new reviewed research package with its own checksum and audit record. A defensible graph edge requires an existing Driver, direction, strength, lag, confidence, and Source/Evidence that supports the mechanism. The outcome series alone cannot establish that causal chain. For item 82, importance weight 0.8 is generated by indicator-importance-rubric-v1. Its five committed components are directness 0.85, population reach 0.7, persistence 0.8, effect on human capabilities 0.8, and overlap control 0.85. The formula weights them 0.35, 0.25, 0.20, 0.15, and 0.05, then rounds to the nearest 0.05. The value is deterministic and reproducible; it is not an intuitive score assigned while drafting prose. The 2018-2024 sequence for item 82 is too short for strong causal or structural claims. Cross-economy comparability can be affected by institutional definitions, survey coverage, modelled estimates, missing observations, classification practices, revisions, and unequal statistical capacity. A global aggregate or unweighted reporting-economy mean can also conceal distributional extremes and should not be used as an individual-level estimate. The item 82 NormalLine is a transparent comparative reference, not a claim that every jurisdiction has the same legal target or feasible frontier. The source may revise both historical points and metadata. Any later refresh must preserve the earlier Source, Evidence, checksum, and audit trail; it must not mutate the meaning of an already reviewed series identity. Historical extension (added 2026-10): V-Dem via World Bank Data360, 1900–2017, transform none.