Measured Minds — tested cognitive performance by country
Three datasets, one honest picture: the contested NIQ "national IQ" compilation shown next to two rigorous measured assessments — HLO World Bank Harmonized Learning Outcomes and PISA 2022 — all rescaled to a common IQ-style index (100 = anchor, SD 15).
Read this first: test performance is not innate ability. Every serious analysis of these datasets shows scores tracking schooling access, nutrition, health, and wealth — and rising fast where those improve (the Flynn effect: ~3 points/decade during development). The "national IQ" (NIQ) layer is included because it circulates widely, not because it is reliable: where it disagrees with measured data, trust the measured data. The scatter tab shows exactly where.
—countries shown
—with measured data
—with 2+ sources
—NIQ-only (treat with caution)
Methodology, honest caveats & what this does not claim
The common scale. PISA and HLO scores (≈300–625, design mean 500/SD 100) are rescaled as index = 100 + (score − 500) × 0.15. NIQ is already on an IQ-style metric. The composite is the plain average of whichever sources exist for a country. The anchors differ (PISA/HLO ≈ OECD-participant mean; NIQ ≈ UK norm), so treat the index as a relative comparison, never a clinical measurement of any person.
Why NIQ is flagged. The Lynn/Becker compilation is widely criticized in the psychometrics literature: many national values are imputed from neighboring countries (never measured), and several "measured" values rest on small, unrepresentative, or decades-old samples. Countries whose only data is NIQ are drawn faded and, where thin or imputed, hatched on the map and flagged in every view. Even among the 130 "measured" NIQ countries, 47 rest on just 1–2 samples (Mongolia: one) — sample counts are shown per country. Becker's own manual flags Nicaragua's value (60.22) as a likely data error; it is footnoted in its tooltip.
What this does not claim. It does not rank the intelligence of peoples. It does not claim group differences are innate or fixed — the measured datasets themselves show large within-country gaps by school access and income, and rapid national gains over decades, which no genetic explanation fits. It is a picture of tested performance under very unequal conditions.
Composite index by country. Brighter = higher tested performance. Solid fill = measured data (HLO and/or PISA); faded + hatched = the only number is the contested NIQ estimate — often imputed, shown so the gap in real data is visible rather than hidden. Click any country for its full record.
100+ top tier93–99.986–92.978–85.9<78NIQ-only, 3+ samplesNIQ-only, thin (≤2 samples)NIQ-only, imputedno data
Ranked composite. Every bar carries its source chips — a rank built on measured data is not the same as a rank built on an imputed guess. NIQ-only rows are faded.
What NIQ claims vs what measurement finds. Each dot is a country with both an NIQ value and measured data. On the diagonal, they agree. Orange = NIQ claims lower than measured performance supports; blue = NIQ claims higher. Systematic orange in low-income countries is the documented downward bias of the NIQ dataset.
The full record. Click headers to sort. Basis column: measured = real assessment in that country; imputed = estimated from neighbors, never measured.