The Civica Index methodology.

Version v1.0·Apr 2026·DOI pending

The Civica Index is a composite governance score assigned to every sovereign state — a structural score (CI) updated quarterly, and a real-time Pulse (CP) updated daily from classified events. Transparent sources. Fixed weights. Reproducible. Citable.

Section 1Purpose

The Civica Index measures overall quality of governance — democratic participation, institutional strength, rule of law, human development, freedom, and security — through two complementary scores:

  • Civica Index (CI)— structural, updated quarterly; answers “how well-governed over time?”
  • Civica Pulse (CP)— event-sensitive, updated daily; answers “what’s the state right now?”

The Index is transparent (every component traces to a public source), reproducible (methodology fully documented), and citable (structured for academic, journalistic, and institutional use).

Section 2The 0–100 scale

Both CI and CP are expressed on a 0–100 scale, where higher is better. Five interpretive tiers:

Section 3Civica Index — structural

3.1 · The six dimensions

The CI is a weighted composite of six dimensions, each sourced from established, peer-reviewed or institutionally maintained open datasets. Raw values are normalized to 0–100 before weighting:

DimensionWeightPrimary sourceSecondary source
Democratic quality30%V-Dem Liberal DemocracyV-Dem Polyarchy
Rule of law & institutions20%World Bank WGIV-Dem Rule of Law
Human development15%UNDP HDIWB income/edu/health
Freedom & rights15%Freedom in the WorldRSF Press Freedom
Corruption control10%Transparency Int'l CPIWGI Control of Corruption
Stability & security10%Global Peace IndexFragile States Index

3.2 · Normalization

Each source uses a different native scale. All are normalized to 0–100 using min-max normalization across the full dataset of countries:

normalized_score = ((raw_value − global_min) / (global_max − global_min)) × 100

For inverted scales (where lower = better, such as GPI and Freedom House), the normalization is reversed.

3.3 · Composite

CI = (0.30 × democratic_quality)
   + (0.20 × rule_of_law)
   + (0.15 × human_development)
   + (0.15 × freedom_rights)
   + (0.10 × corruption_control)
   + (0.10 × stability_security)

3.4 · Update frequency

The CI updates quarterly. Between publications, the CI carries forward the most recent available value per source. The CI never moves between quarterly recalculations — that’s what the Pulse is for.

3.5 · Missing data

If a country has no data for a dimension, that dimension is excluded and remaining weights are re-proportioned to sum to 100%. The CI is flagged “partial”. If fewer than three of the six dimensions have data, no CI is calculated.

Section 4Civica Pulse — real-time

4.1 · Concept

The Pulse starts at the country’s current CI and is modified by a real-time event impact layer:

CP = CI + EventImpact

4.2 · Event ingestion

An automated pipeline monitors global news sources daily (GDELT, Google News, Reuters/AP, country gazettes). It filters for events tagged to a specific country that fall within governance-relevant categories.

4.3 · Event classification

Each qualifying event is classified by an LLM agent into one of twelve directional categories — from armed conflict (negative) to democratic election (positive). Every score is logged with category, severity, confidence, and a one-sentence justification. All are published in the Pulse changelog for audit.

4.4 · Severity scale (−10 to +10)

ScoreMeaningExample
−10Catastrophic governance failureFull-scale war, genocide, state collapse
−7 to −9SevereMilitary coup, mass civilian casualties, total media blackout
−4 to −6SignificantMajor protest crackdown, opposition leader imprisoned
−1 to −3ModerateJournalist arrested, minor corruption scandal
0NeutralRoutine diplomatic event, sports, weather
+1 to +3Moderate positiveMinor reform, transparency improvement
+4 to +6Significant positiveAnti-corruption conviction, rights expansion
+7 to +10Transformative positivePeaceful democratic transition, comprehensive peace agreement

4.5 · Temporal decay

Events lose influence over time. The decay is exponential with a 30-day half-life:

decayed_impact = severity × confidence × e^(−0.693 × days / 30)

An event from today counts 100%. From 30 days ago, 50%. From 60 days, 25%. From 90 days, 12.5%. From 120+ days, negligible (<6%).

4.6 · Aggregation and clamp

EventImpact = Σ (severity × confidence × e^(−0.693 × days / 30))
CP          = clamp(CI + clamp(EventImpact, −30, +30), 0, 100)

The ±30-point cap on EventImpact prevents a single catastrophic event from completely overriding years of structural data.

4.7 · Update frequency

The Pulse recomputes daily. Every Pulse movement comes with a public changelog entry showing every contributing event, its category, severity, confidence, and justification.

Section 5Government-type modifier

The CI does not apply a fixed bonus or penalty per government type. Instead, the site publishes an empirical observation layer — Governance Outcomes by Government Type — showing average CI, distribution spread, and 20-year trajectories for each category. The data speaks for itself.

Section 6Government taxonomy layers

Civica now keeps three separate government labelsfor each country so the site can stay academically honest without flattening unlike systems into one bucket.

  • Raw source label. The CIA World Factbook wording is preserved exactly for provenance.
  • Normalized regime type. Bjornskov-Rode / CGV coding tracks executive-legislative accountability in the parliamentary / semi-presidential / presidential democracy and civilian / military / royal dictatorship families.
  • Derived structural form.Civica derives the system's constitutional form from explicit primitives such as federalism, monarchy, executive structure, and government dependency, with documented overrides for edge cases.

These layers are classification metadata only. They do not change CI weights, dimension scores, or the composite formula. Their purpose is interpretive: to help readers compare systems without mistaking structural form for a scoring bonus.

Switzerland is a useful example of why the layers stay separate: BR/CGV codes it as a presidential democracyin the accountability sense, while Civica's structural layer exposes itsfederal directorial republic form. Both statements are true, and the disagreement is informative rather than an error.

Section 7Data quality & limitations

Known limitations. The CI is only as current as its slowest-updating source (some indices publish 12–18 months behind). Pulse event scoring relies on LLM judgment, which may exhibit biases in event selection or severity. Countries with limited media coverage will have fewer detected events, potentially making their Pulse artificially stable. The ±30-point cap prevents extreme scores but may understate truly catastrophic situations.

Mitigations. All LLM event scores are logged with justifications for audit. A quarterly human review samples 5% of scored events. The methodology is versioned; weight/source/formula changes publish with a changelog. Countries with fewer than 3 events in 90 days are flagged low-confidence.

Section 8Publication & citation

Civica Index 2026. Civica Atlas. https://civicaatlas.org/civica-index
Civica Pulse for [Country], [Date]. Civica Atlas. https://civicaatlas.org/civica-index/[slug]

API access

GET https://civicaatlas.org/api/v1/index/{country_slug}
GET https://civicaatlas.org/api/v1/pulse/{country_slug}
GET https://civicaatlas.org/api/v1/index/rankings
GET https://civicaatlas.org/api/v1/pulse/changelog/{country_slug}

A small embeddable badge showing a country’s CI and CP is available for news organizations, blogs, and other platforms to embed with attribution.

Section 9Site integration

PageWhat it shows
/civica-indexGlobal leaderboard · all countries ranked by CI or CP, filterable
/civica-index/[slug]Country detail · dimension breakdown, Pulse changelog, history
/civica-index/compareOverlay two or three countries on a timeline
/civica-index/methodologyThis document
/civica-index/government-typesGovernance outcomes by government type
/civica-index/changelogGlobal feed of all Pulse movements across all countries

Section 10Versioning

Version
v1.0
Published
Apr 2026
Countries covered
197
Open-source datasets
9

Changes to weights, sources, formulas, or the event-scoring framework are documented in a public changelog and assigned a new version number. Historical CI/CP values are never retroactively recalculated; they’re preserved as-is with their methodology version recorded. This way, cited values remain stable forever.

Changelog

  • vv1.0 — Apr 2026 · Initial CI methodology — 6 dimensions, min-max normalization