Back to home

Methodology

How we measure trust—transparently, with open data only.

🚧

This is a Beta Release

Trust Atlas is under active development. Data coverage is incomplete, methodologies may change, and you'll likely find rough edges. We're building in the open because we believe transparency matters—even when things aren't perfect.

Found an error? Have a suggestion? Want to contribute data or code?

Open an issue on GitHub

Overview

Trust Atlas measures trust across three independent dimensions. Each pillar is displayed separately—we do not combine them into a single composite score. All data sources are freely accessible—no paywalls, no proprietary datasets.

Why no composite index? Survey data is sparse (WVS every ~5 years) while governance data is annual. Combining them creates artificial volatility. Individual pillars tell clearer stories.

The Three Pillars

🤝

Interpersonal Trust

Do people trust each other? Based on the classic survey question: “Generally speaking, would you say that most people can be trusted, or that you need to be very careful in dealing with people?”

Sources: WVS, GSS, ANES, CES
WVS-family methodology: We use only surveys that share identical or highly comparable question wording and response scales. The World Values Survey, General Social Survey, American National Election Studies, and Canadian Election Study all use the same binary “trust/careful” question established by Rosenberg (1956). Other surveys (ESS, regional barometers) use different scales that produce systematically different scores, so we exclude them to ensure methodological consistency.
🏛️

Institutional Trust

Do people trust their government and institutions? Measures confidence in national government, parliament, courts, and other public institutions.

Sources: WVS, GSS, ANES, CES
Same methodology rationale: Like interpersonal trust, we limit institutional trust to WVS-family surveys. Regional barometers, ESS, OECD, and others use different question wording, response scales, and institution definitions that produce incompatible scores. For example, Asian Barometer Wave 1-2 used inverted scales producing 10% scores vs. 95% in later waves. Mixing methodologies creates artificial volatility in time series.
⚖️

Governance Quality

How trustworthy are institutions objectively? Expert assessments of corruption, rule of law, and government effectiveness serve as a proxy for institutional integrity.

Sources: Transparency International CPI, World Bank WGI, WJP Rule of Law Index, Freedom House, V-Dem
Governance = 0.5 × CPI + 0.5 × WGIrescaled

Why Independent Pillars?

Many trust indices combine multiple measures into a single composite score. We deliberately chose not to. Here's why:

Different Data Rhythms

Survey data (WVS, GSS) is collected every 5-7 years. Governance data (CPI, WGI) is annual. Combining them creates artificial volatility—a country's “trust score” would swing wildly based on which data source updated most recently.

Different Constructs

“Do you trust your neighbor?” and “Is the government corrupt?” measure fundamentally different things. Research shows a “trust paradox”—some societies show high interpersonal trust despite low governance scores. Averaging them obscures this insight.

Clearer Stories

A country with high interpersonal trust but low institutional trust tells a different story than one with the opposite pattern. Separate pillars let researchers and journalists see these nuances instead of burying them in a single number.

Data Confidence

Not all data is equally reliable. We show a confidence indicator and “likely range” for each score so you know how much weight to give it.

High confidence

Recent survey data (within 3 years). This is our best data.

Likely range: ±5 points
Moderate confidence

Older survey data (3-7 years). Still useful, but conditions may have changed.

Likely range: ±10 points
Estimate

Based on governance indicators only—no direct survey data available.

Likely range: ±15 points
What's a “likely range”? It's the window where the true value probably falls. If Russia shows 24% with a range of 19-29, the actual level of interpersonal trust is most likely somewhere in that band. When comparing countries, overlapping ranges mean we can't confidently say which is higher.

Normalization

All scores are normalized to a 0-100 scale for easy comparison:

  • Survey percentages (WVS, GSS, etc.) — used directly (already 0-100)
  • CPI scores — used directly (already 0-100)
  • WGI scores (-2.5 to +2.5) — rescaled: ((x + 2.5) / 5) × 100
  • WJP Rule of Law Index — used directly (already 0-100)

Data Sources

We collect data from many sources but apply strict methodology rules for the three pillars shown on the Explore page. Other sources remain available for stories and deeper analysis.

Used for Pillar Scores

These sources power the Explore page. We use only methodologically compatible surveys to ensure scores are comparable across countries and time.

Interpersonal & Institutional Trust

WVS-family surveys only—identical question wording and response scales:

  • World Values Survey (WVS) — 108 countries, 1981-2023
  • General Social Survey (GSS) — USA, 1972-2024
  • American National Election Studies (ANES) — USA, 1958-2024
  • Canadian Election Study (CES) — Canada, 2008-2021
Governance Quality
  • Transparency International CPI — 187 countries, 2008-2024
  • World Bank WGI — 207 countries, 2008-2023

Additional Sources (Stories & Analysis)

These sources aren't used for pillar scores due to methodology differences, but we use them for featured stories and deeper analysis where single-source consistency matters more than cross-country comparability.

Governance & Rule of Law
  • World Justice Project Rule of Law Index — 142 countries, 2015-2024 (powers “Against the Tide” story)
  • Freedom House — 189 countries
  • V-Dem — 176 countries
Regional Surveys

Different scales/questions than WVS, but valuable for regional deep-dives:

  • Afrobarometer — 34 African countries
  • Arab Barometer — 16 MENA countries
  • Asian Barometer — 18 Asian countries
  • Latinobarómetro — 18 Latin American countries
  • LAPOP AmericasBarometer — 34 countries
  • European Social Survey (ESS) — 30+ European countries
Not Available (Proprietary)

We cannot redistribute data from these sources:

  • Edelman Trust Barometer — restricted redistribution
  • Gallup World Poll — requires paid subscription

API Access

All Trust Atlas data is available via our public API:

GET /api/countries — List all countries
GET /api/country/:iso3 — Time series for a country
GET /api/trends/global?pillar= — Latest scores by pillar (for map)
GET /api/trends/regions?pillar= — Regional averages
GET /api/trends/country/:iso3 — Country trends
GET /api/stats — Database statistics
GET /api/methodology — Methodology as JSON

Source code available on GitHub.

Methodology version 0.3.0 — Last updated December 2025

v0.3.0: WVS-family methodology—limited survey pillars to sources with identical question wording (WVS, GSS, ANES, CES)