Methodology

Each of the 17 hostility metrics is expressed on a 0 to 100 scale where 0 means rare or no records and 100 means severe, documented hostility. Scores come from one of two places:

• •Real sources: ingested from the sources below. Carry a small green “real” chip in the map sidebar and a full citation on the country profile page.
• •Seeded baselines: derived from regional templates and published research where a direct source has not yet been ingested. Clearly marked as placeholder and labeled as such in each country profile.

Scores are never used to rank countries. A difference of five points is within the noise floor of most underlying sources; a difference of twenty points is a signal worth investigating.

The atlas currently draws real data from seven sources totaling 595 verified metric points across 160 countries. Each source below lists what it provides, how it maps to our metric grid, and the specific caveats readers should hold in mind.

Not every metric has a real source behind it yet. These 11 vectors currently use regional templates and deterministic seeding until the sources below are ingested. Nothing is hidden: every seeded metric is flagged as such in the map sidebar and the country profile.

• Ethnic conflict in Africaneeds ACLED country-level aggregation
• Anti-Indigenous biasneeds Indigenous Navigator / MRG aggregation
• Anti-Roma biasneeds FRA EU Fundamental Rights survey data
• Anti-migrant biasneeds IOM and UNHCR cross-tabulation
• Class and caste biasneeds NCRB India + IDSN + Minorities at Risk aggregation
• Ableism (anti-disability bias)needs World Bank Disability Data Initiative
• Ageism and generational conflictneeds ESS + WVS age-cohort data
• Sports rivalry violenceneeds UEFA + FIFA match-day incident archive
• Online subculture harassmentneeds ADL HEAT + Civic Signal digital harassment data

Confidence is distinct from hostility. A country with high reporting confidence may have high or low recorded hostility. A country with low reporting confidence is often a place where independent reporting is difficult, not a place where everyone gets along.

We derive visibility from the Freedom House state-opacity flag (Not Free lowers confidence) plus source count and source family diversity. Countries that refuse to cooperate with international monitors are visualized with a cross-hatch data-void pattern on the map; that silence is a measurable signal, not a gap to fill with assumptions.

When official crime statistics show low hate-crime rates but survey data shows high experienced discrimination, we surface both perspectives with context about underreporting and definitional differences. We never flatten complexity to produce a simpler narrative.

For the United States: the FBI UCR 2022 report shows 11,634 single and multi-bias incidents, but community organizations have consistently documented much higher totals. The FBI number is what we have; the gap is what you should see.

• ×Predict what a specific person believes
• ×Rank countries or groups as “most hateful” without context
• ×Use gamified scoring on painful material
• ×Present harm without linking to bridges and reform stories
• ×Claim to have real data where we have only seeded baselines