Cross-national differences in average cognitive test performance are among the most consistently reported — and most contentiously interpreted — findings in intelligence research. The data showing that measured IQ scores differ between countries are real; what remains deeply contested is what those differences mean, how reliable the underlying measurements are, and what causes them. This article attempts to separate the empirical findings from the interpretive claims, and to present the limitations of this literature with the seriousness they deserve.
The Data: What Is Actually Measured
The most widely cited compilation of national IQ estimates is Richard Lynn and Tatu Vanhanen's database, published in books including IQ and the Wealth of Nations (2002) and subsequent revisions. Lynn and Vanhanen aggregated cognitive test score data from dozens of countries and converted them to an IQ scale normed on a British reference sample. Their data shows substantial variation across countries, with East Asian nations (Japan, South Korea, China) clustered at the high end (105–108), Western European and North American nations in the mid-90s to low-100s range, and sub-Saharan African nations typically estimated in the 70s.
These figures have been widely cited in popular and academic contexts. They should be treated with substantial caution for several reasons. First, the underlying studies vary enormously in quality, sample size, sampling methodology, and test validity. Many national estimates in Lynn and Vanhanen's database rest on a single study with small, unrepresentative samples — a few hundred urban schoolchildren, for instance, cannot represent the cognitive performance of a country of hundreds of millions with vast rural-urban and socioeconomic diversity. The late epidemiologist Richard Pfeifer documented numerous errors, questionable data inclusions, and methodological inconsistencies in the Lynn dataset. A competing dataset compiled by Becker (2019) uses more stringent inclusion criteria and produces somewhat different — though still heterogeneous — national estimates.
The Overwhelming Role of Environmental Factors
The scientific consensus on why national differences in measured IQ exist points overwhelmingly to environmental, not genetic, factors. The strongest evidence comes from within-country trends over time (the Flynn Effect), natural experiments in nutrition and health, and cross-national comparisons that control for socioeconomic variables.
Nations with low average test scores are, almost without exception, nations with high rates of childhood malnutrition, high infectious disease burden, limited access to formal education, high rates of iodine and iron deficiency, significant lead and toxin exposure, and limited access to the abstract problem-solving environments that IQ tests are specifically designed to assess. Each of these factors, individually, has been shown in well-controlled studies to suppress cognitive performance. Cumulatively, their effects on national averages can be very large — potentially accounting for much or all of observed national differences.
The Flynn Effect provides the most powerful argument against genetic interpretations of national differences. In Ireland, the average IQ rose by approximately 14 points between 1972 and 2002 — a shift too fast to be explained by genetic change and directly attributable to improvements in nutrition, education, and living standards. Similar trajectories have been documented in South Korea, Japan (post-war), and Estonia (post-Soviet). Countries that were once in the "low" tier have moved substantially toward the global mean as living conditions improved, demonstrating that the performance differences are not fixed features of populations but reflections of current conditions.
The Causal Claims and Why They Are Premature
Despite the weak evidentiary basis for genetic explanations of national differences, some researchers have used the Lynn dataset to argue for heritable differences in cognitive capacity between populations. These arguments are premature for several reasons beyond the data quality problems noted above. First, establishing that individual-level IQ is substantially heritable within a population does not imply that differences between populations are genetic in origin — this is a well-recognised logical fallacy, analogous to noting that plant height is heritable while failing to recognise that a plant grown in poor soil will be shorter regardless of its genetic potential. Second, we have no scientific basis for estimating what the "genetically expected" IQ of any national or ethnic population would be in optimised environmental conditions, making between-group genetic comparisons methodologically incoherent with current tools. Third, the populations we observe have been shaped by millennia of different environmental histories — disentangling genetic from environmental causation at the population level is, with current methods, essentially impossible.
Key Takeaway
Cross-national differences in measured cognitive test performance are real but unreliable in their specifics, and they are overwhelmingly explained by differences in the environmental conditions — nutrition, health, education quality, toxin exposure — that either enable or suppress the expression of human cognitive potential. The available evidence provides no scientifically credible basis for attributing these differences to genetic differences between national or ethnic populations. The practically important conclusion is not about the fixed cognitive characteristics of any group, but about the modifiability of cognitive outcomes: the same environmental improvements that produced 10-15 point IQ gains in Ireland, Japan, and South Korea over recent decades are available, in principle, to every population currently underperforming its potential due to preventable deprivation.