IQ

The most robust finding about IQ is its predictive validity for outcomes that involve cognitive processing. IQ predicts educational attainment with correlations around 0.5. It predicts job performance with correlations around 0.3-0.5, varying by job complexity — higher for complex jobs, lower for routine ones. It predicts income modestly, longevity modestly, health behaviors modestly, and even specific life events like divorce and incarceration. The prediction is probabilistic and applies at the population level; individual outcomes depend on many factors, and IQ is one among several. But the prediction is stable and substantial enough to be one of the most replicated findings in behavioral science.

The heritability of IQ is substantial. Twin and adoption studies consistently estimate that roughly fifty to eighty percent of the variance in adult IQ within a population is attributable to genetic factors, with higher estimates in adulthood than in childhood. This heritability does not mean IQ is unchangeable — heritability refers to the source of variation within a specific environmental range, not to the fixedness of the trait — but it does mean that the biological component is real and substantial. Environmental influences matter but produce smaller effects than naive environmentalists assumed.

Group differences in average IQ scores are the most contested area. Differences between racial and ethnic groups, between men and women on specific subtests, and between nations have been documented. Interpretation is where the public discussion becomes explosive. Some researchers argue the differences reflect biological factors; others argue they reflect environmental, educational, and measurement factors. The evidence is mixed and the political stakes enormous. The academic consensus is that population-level differences exist in the measurements, that their causes involve substantial environmental components, and that the findings do not support hereditarian claims about racial hierarchy.

The practical use of IQ tests in high-stakes decisions — educational admission, employment, military selection — has produced both documented utility and documented harm. Well-validated tests predict performance in cognitively demanding roles better than most alternatives. Poorly-designed or poorly-applied tests have excluded capable candidates on the basis of cultural biases or subtests that do not actually predict the target outcome. The appropriate use is a practical question with specific trade-offs.

IQ persists as a topic because the underlying phenomenon is real and consequential. Cognitive ability varies across individuals, and the variation matters for outcomes people care about — success in demanding educational programs, effectiveness in complex jobs, ability to learn new skills. Abandoning the measurement does not make the variation disappear; it just means decisions that implicitly account for cognitive ability do so without explicit measurement. The topic is uncomfortable, but the thing it points at does not go away when the pointing is suppressed.

The topic also persists because every few years someone rediscovers the research and writes a book that re-ignites the debate — Herrnstein and Murray's The Bell Curve, Charles Murray's later work, Steven Pinker's discussions, and more recently Substack writers and podcasters taking up the subject. Each rediscovery produces the same pattern: public condemnation, defensive backlash, careful researchers trying to explain what the evidence shows, and the topic retreating to academic obscurity until the next rediscovery.

IQ matters because it is one of the best-validated and most-researched measurements in psychology, and because ignoring what it measures does not make the measured phenomenon go away. Decisions about education, employment, and social policy are made every day that implicitly account for cognitive ability. Whether these decisions are made with explicit measurement and conscious thought, or implicitly without either, produces different outcomes. The explicit version has specific failure modes but also specific advantages; the implicit version has specific failure modes without the advantages. Which version prevails depends partly on whether the topic can be discussed carefully enough to make the explicit version viable.