FOLK
CULTURAL INTELLIGENCE
FOLK Methodology Note

How the FOLK Scores Were Built

Methodology note: the four orientations, the statistical baseline, and the AI-council scoring

The scores were produced in three stages: a statistical stage that defined the four orientations and a starting baseline number for each country; an AI council stage that reviewed evidence and qualitative literature; and an integrator stage that set the final published score.

Part 1

Defining the Four Orientations (Statistics)

Five established cultural frameworks were used as inputs: Hofstede, Trompenaars, GLOBE, Schwartz, and the World Values Survey (WVS). Supplying thirty-four cultural indicators across ~25 complete-data countries, factor analysis (Multiple Factor Analysis) was run to find how many distinct underlying orientations were present. It empirically identified four.

The name FOLK encodes Four Orientations of Life and Kinship:

  • Kinship (Who you belong to & show it)Identity (Social → Self): Where identity sits — group vs. individual.
    Expression (Restrained → Open): How openly warmth and emotion are displayed.
  • Life (How you organize & strive)Structure (Fluid → Certain): Preference for rules vs. flexibility.
    Drive (Accepting → Striving): Contentment vs. competitive achievement.

Sample Expansion & MICE Imputation

Framework Coverage LevelCountries Scored
All five frameworks~27 countries
At least four60 countries
At least three84 countries
At least two120 countries
At least one171 countries
No framework data (Extension)26 countries (Direct AI Council Evaluation)

Missing values for 171 countries with partial framework data were filled via MICE (Multivariate Imputation by Chained Equations, 40 draws). Stability was validated using Tucker's φ (0.95+ green). The remaining 26 countries with no framework data were evaluated directly by the AI Council from qualitative evidence.

Empirical Factor Structure

FOLK Factor Structure Matrix (36 Indicators)

Cross-framework factor loadings across Hofstede (HOF), GLOBE (GLO), Schwartz (SCH), World Values Survey (WVS), and Trompenaars (TRO).

FWAttribute / IndicatorD1 (Identity)D2 (Expression)D3 (Structure)D4 (Drive)AssignedNotes & Mechanism
WVSChoice0.959D1Self pole
WVSEquality0.915D1Self pole
GLOGLOBE UA — conformity pressure-0.904D1Social pole (Empirical surprise)
SCHEmbeddedness-0.892D1Social pole
HOFIndividualism0.872D1Self pole
HOFPower Distance-0.872D1Social pole
SCHAffective Autonomy0.854D1Self pole
SCHIntellectual Autonomy0.841D1Self pole
TROAscription-Achievement0.841D1Self pole
TROParticularism-Universalism0.775D1Self pole
TRODiffuse-Specific0.756D1Self pole
WVSVoice0.6750.461D1Also Open: Voice = self-expression & open communication
SCHHierarchy-0.646D1Social pole
TROCommunitarianism-Individualism0.640D1Self pole
SCHEgalitarianism0.5910.683D1Also Open: flat social relations produce autonomy & openness
GLOGender Egalitarianism0.5390.636D1Also Open: flat gender norms
WVSDefiance0.521D1Self pole
WVSAutonomy0.501D1Self pole
GLOInstitutional Collectivism-0.5100.568D1Social pole & Open (in-group warmth)
HOFLong-Term Orientation-0.807D2Restrained pole
HOFIndulgence0.797D2Open pole
TROFuture-Past Orientation-0.811-0.807D2Past-oriented = Social & Restrained
GLOIn-Group Collectivism0.625D2Open pole
GLOPower Distance (practices)-0.547-0.620D2Restrained pole & Accepting
WVSScepticism0.490D2Open pole
GLOHumane Orientation0.455D2Open pole
GLOAssertiveness-0.494-0.573D2Restrained pole & Fluid
HOFUncertainty Avoidance0.8580.333D3Certain pole & Striving
SCHHarmony0.589D3Certain pole
GLOFuture Orientation0.420D3Reassigned from D2 (Planning = managing uncertainty)
SCHMastery0.680D4Striving pole
WVSRelativism-0.644D4Accepting pole
GLOPerformance Orientation0.425D4Reassigned from D2 (Achievement = Drive)
HOFMasculinity0.231D4Striving pole
TROAffective-Neutral0.415-0.510CrossSpreads across D1 & D2
TROExternal-Internal0.5500.420CrossSpreads across D1 & D2 (Internal control = Self & Open)
Part 2

The Baseline Score & Remaining Statistical Issues

Statistical baseline scores were accompanied by confidence intervals (e.g. Bhutan Identity: baseline 45, interval 28.6–57.2). However, raw statistical scores suffered from three limitations:

1. Scale CompressionScores clustered heavily near mid-scale (50), leaving much of the 0–100 range under-utilized.
2. Identical ScoresMultiple countries received identical imputed values across certain orientations.
3. Flat ProfilesHeavy imputation created flat profiles across orientations for data-sparse countries.
Part 3

Multi-Agent AI Council Architecture

To prevent single-model confirmation bias, the council runs six specialized expert roles across multiple underlying language models (Claude, OpenAI, DeepSeek):

Cultural Anthropologist

Kinship, ritual, custom, and everyday practice.

Institutional Analyst

The state, law, religion, and how rules and power operate.

Statistician

Keeps the score consistent with quantitative data and baseline.

Comparativist

Checks the score against regional peer groups and similar countries.

Country Specialist

Supplies qualitative nuance and country-specific detail.

Skeptic

Argues against the emerging answer and finds the weakest point.

Part 4

Formulaic Score Pipeline to Final Decision

1. Baseline Reference

MFA Factor Baseline

2. Council Consensus

Weighted Vote Average

3. Recommended Score

Formulaic & Reproducible

4. Final Score

Integrator LLM Decision

SUMMARY

34 Indicators • 197 Countries • 1 Model

Factor analysis over 34 indicators defined the baseline; six specialized AI council agents reviewed qualitative evidence; and an Integrator set the final score.

Explore Countries