The Science of IQ

1 · What IQ tests measure

In 1904, Charles Spearman noticed something that remains one of the most replicated findings in psychology: people who do well on one kind of mental task tend to do well on all kinds of mental tasks. Vocabulary, mental arithmetic, spatial puzzles, reaction time - performance on every cognitive test correlates positively with every other. Spearman called the statistical factor underlying this "positive manifold" g, for general intelligence.[1]

A century of factor-analytic work has refined, not overturned, that picture. John Carroll's landmark reanalysis of more than 460 datasets produced the three-stratum model: narrow abilities (stratum I) group into broad abilities like fluid reasoning, verbal comprehension, spatial processing, memory, and processing speed (stratum II), which in turn all load on g at the top (stratum III).[2] The modern synthesis of this work - the Cattell–Horn–Carroll (CHC) model - is the blueprint behind today's major tests.[3]

Two broad abilities deserve special mention:

• Fluid intelligence (Gf) - solving novel problems with no learned answer: spotting the rule in a pattern matrix, holding and transforming information. It peaks in early adulthood and declines gradually with age.[4]
• Crystallized intelligence (Gc) - accumulated knowledge and vocabulary. It keeps rising through middle age and is far more resistant to aging.[4]

An IQ score is not a count of anything physical - it is a rank on the population, rescaled so the mean is 100 and the standard deviation is 15. About 68% of people score between 85 and 115, and roughly 2% score above 130.[5] g itself is remarkably robust: a general factor emerges from any sufficiently diverse battery of cognitive tasks, and the g factors extracted from different batteries correlate near unity.[6]

2 · How real tests are built and scored

Professional instruments like the Wechsler Adult Intelligence Scale (WAIS) or the Raven's Progressive Matrices go through a construction pipeline that is easy to underestimate:

• Item analysis. Hundreds of candidate items are trialed; items are kept based on difficulty, discrimination, and freedom from bias, often using item response theory.
• Norming. The test is administered to a large sample carefully stratified to match the census on age, sex, education, region, and ethnicity - thousands of participants. Your score means "where you stand relative to this reference sample."[5]
• Reliability. Full-scale IQ from a major battery has internal consistency around .97–.98 and test–retest reliability around .95, giving a standard error of measurement of roughly 2–3 points.[5] Even so, a single sitting reports a confidence band, not a point.
• Validity. Scores must predict external criteria (school achievement, training success) and correlate appropriately with other established tests.[4, 7]

Raven-style pattern matrices are a staple of research because they are among the most g-loaded single formats known, require no language, and travel well across cultures - which is why matrix reasoning anchors our tests too.[8]

Online tests - including ours - cannot fully replicate this pipeline: there is no proctor, no census-matched norming sample, and far fewer items. A well-built online test can still rank reasoners meaningfully, but its precision is lower and its norms are provisional. We explain exactly how we handle this on the methodology page.

3 · Genes and environment

Intelligence runs in families, and a huge body of twin, adoption, and (recently) DNA-based research has quantified why. The consistent findings:[9, 10]

• Heritability is substantial - and grows with age. Genetic differences account for roughly 20% of IQ variation in infancy, ~40–50% in childhood, and 60% or more in adulthood - the "Wilson effect."[10, 11] Identical twins reared apart end up remarkably similar in IQ; adopted siblings reared together end up barely correlated as adults.[11]
• No single "intelligence gene" exists. Genome-wide studies find thousands of variants, each with a vanishingly small effect; polygenic scores currently explain only a modest slice (~4–10%) of IQ variance.[10]
• Environment matters most where it is worst. Severe deprivation, malnutrition, iodine deficiency, and lead exposure all depress cognitive development; adoption from deprived into enriched homes raises IQ substantially.[9, 12] Some US samples show heritability is lower in low-SES families, though this moderation replicates inconsistently outside the US.[13]

The key misreading to avoid: "heritable" does not mean "fixed." Height is ~80–90% heritable, yet average height rose dramatically in the 20th century as nutrition improved. The Flynn effect (below) proves average IQ responds to environment on a large scale, at the very same time that individual differences within a generation are substantially genetic.[9, 14]

4 · What IQ predicts - and how strongly

IQ is among the best-validated predictors in psychology, but "best in psychology" still means probabilistic, not destiny. Representative effect sizes from meta-analyses and large cohorts:

• School achievement. Cognitive ability measured at age 11 correlates about .8 (at the latent level) with national exam results at 16 - the strongest single predictor known.[15]
• Job performance. General mental ability predicts performance across occupations (validity ≈ .3–.5, higher in more complex jobs), outperforming interviews, references, and personality inventories in classic meta-analyses.[7]
• Socioeconomic outcomes. Meta-analytically, IQ correlates ≈ .56 with educational attainment, ≈ .45 with occupational status, and ≈ .23 with income - real, but leaving most variance to other factors.[16]
• Health and longevity. In a meta-analysis following over a million people, each 1-SD (15-point) advantage in youth IQ was associated with a 24% lower risk of death over follow-up - a field now called cognitive epidemiology.[17] Mechanisms likely include education, safer jobs, health literacy, and shared bodily "system integrity."[4]

Two honest caveats. First, these are group-level regularities: plenty of high-IQ individuals struggle and plenty of average scorers thrive. Second, conscientiousness, self-control, opportunity, and luck all carry independent predictive weight - intelligence is one ingredient, not the recipe.[9, 16]

5 · The Flynn effect: rising (and falling) averages

James Flynn documented that raw IQ-test performance rose massively across the 20th century - about 3 points per decade in 14 nations, forcing test publishers to re-norm periodically.[14] A 2014 meta-analysis spanning 285 studies and nearly four million participants confirmed the gain at roughly 2.3–3 points per decade, remarkably steady across eras.[18]

Because genes cannot change that fast, the Flynn effect is decisive evidence that environment moves test scores at scale. Leading explanations - better nutrition and health, longer and more abstract schooling, smaller families, and a daily life saturated with symbolic thinking - likely all contribute.[18, 19] Gains are largest on fluid, matrix-style tests and smaller on vocabulary, suggesting people have grown better at abstract problem-solving specifically.[19]

Since the 1990s, several high-income countries (notably in Scandinavia) have seen the effect stall or reverse. Norwegian conscript data show the decline occurs within families - brothers born later score lower - pointing at environmental causes rather than demographic change.[20]

6 · Limits, criticisms, and misuse

An honest site about IQ has to include this section.

• IQ is not the whole mind. Standard tests do not measure creativity, wisdom, curiosity, emotional skill, or rational decision-making - Keith Stanovich's work shows intelligence and rationality can dissociate sharply.[21] The APA task force report and its update remain the best consensus summaries of what tests do and don't capture.[9, 12]
• Motivation moves scores. Offering incentives raises IQ scores by around 0.6 SD in low-stakes settings - so a score reflects effort as well as ability, especially online.[22]
• Test anxiety and stereotype threat. Situational pressure can depress performance for stereotyped groups, though the size and generality of stereotype threat is actively debated, with meta-analyses suggesting smaller effects than early studies implied.[23]
• Historical misuse is real. Early-20th-century testing was entangled with eugenics, forced sterilization, and discriminatory immigration policy. That history is a permanent warning about over-reading test scores, and it is why modern professional standards restrict how scores may be used.[9]
• Group differences are not genetic verdicts. Average score gaps between groups have narrowed over time and are fully compatible with environmental explanations - the same tools that show within-group heritability say nothing automatic about between-group causes.[9, 12]
• One number flattens a profile. Two people with identical full-scale scores can have very different strengths across domains - which is why our results report a domain profile, not just a single figure.

7 · Can you raise your IQ?

The trial literature is clearer than internet folklore suggests:

• Education works. Across natural experiments and policy changes, each additional year of schooling raises IQ by roughly 1–5 points, with effects persisting across the lifespan - the most robust known intervention.[24]
• "Brain training" mostly doesn't. Working-memory and app-based training reliably improves the trained task, but meta-analyses find little or no transfer to untrained reasoning - the "near transfer only" verdict.[25, 26]
• Early-childhood programs boost scores, then fade. Intensive preschool interventions produce real IQ gains that typically shrink after the program ends - though lasting benefits on life outcomes can remain.[12]
• Body and brain. Correcting iodine deficiency, reducing lead exposure, and treating sleep disorders protect cognition; a single bad night of sleep measurably impairs attention and working memory. Aerobic exercise shows modest cognitive benefits, strongest in older adults.[9, 12]

Practical translation: you can absolutely become better at reasoning tasks, sharpen skills, and protect your cognitive health - but no supplement or app has been shown to add lasting points to g in healthy adults. Be skeptical of anyone selling one.

8 · Frequently asked questions

How accurate is an online IQ test?

Less precise than a proctored clinical battery - fewer items, no controlled conditions, provisional norms. Treat any online score (ours included) as an estimate with a band of roughly ±9–15 points, and be wary of sites that promise exact scores or charge for "certified" results.

Does IQ change over a lifetime?

Rank order is surprisingly stable: in the Lothian Birth Cohorts, IQ at age 11 correlated about .66 with IQ at age 77–80.[27] Absolute abilities shift - fluid reasoning declines with age while knowledge grows[4] - and individual scores wobble across sittings.

What score do high-IQ societies require?

Mensa admits the top 2% - approximately IQ 130 on a 15-SD scale - based on supervised tests only. No online test qualifies, ours included.

Will practicing IQ tests raise my score?

Yes, on that test format - retest and practice effects are typically several points[5] - which is one more reason a practiced score overstates ability. Wait weeks between attempts and treat your first sitting as the most informative.

Is IQ just "book smarts"?

No - g emerges from tasks with no schooling content at all (matrices, reaction time, memory span). But see section 6: important human qualities lie entirely outside it.

References

1. Spearman, C. (1904). "General intelligence," objectively determined and measured. American Journal of Psychology, 15(2), 201–292.
2. Carroll, J. B. (1993). Human Cognitive Abilities: A Survey of Factor-Analytic Studies. Cambridge University Press.
3. McGrew, K. S. (2009). CHC theory and the human cognitive abilities project. Intelligence, 37(1), 1–10.
4. Deary, I. J. (2012). Intelligence. Annual Review of Psychology, 63, 453–482.
5. Wechsler, D. (2008). WAIS-IV Technical and Interpretive Manual. Pearson.
6. Johnson, W., Bouchard, T. J., et al. (2004). Just one g: Consistent results from three test batteries. Intelligence, 32(1), 95–107.
7. Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology. Psychological Bulletin, 124(2), 262–274.
8. Raven, J. (2000). The Raven's Progressive Matrices: Change and stability over culture and time. Cognitive Psychology, 41(1), 1–48.
9. Neisser, U., Boodoo, G., Bouchard, T. J., et al. (1996). Intelligence: Knowns and unknowns. American Psychologist, 51(2), 77–101.
10. Plomin, R., & von Stumm, S. (2018). The new genetics of intelligence. Nature Reviews Genetics, 19, 148–159.
11. Bouchard, T. J. (2013). The Wilson effect: The increase in heritability of IQ with age. Twin Research and Human Genetics, 16(5), 923–930.
12. Nisbett, R. E., Aronson, J., Blair, C., et al. (2012). Intelligence: New findings and theoretical developments. American Psychologist, 67(2), 130–159.
13. Tucker-Drob, E. M., & Bates, T. C. (2016). Large cross-national differences in gene × socioeconomic status interaction on intelligence. Psychological Science, 27(2), 138–149.
14. Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ tests really measure. Psychological Bulletin, 101(2), 171–191.
15. Deary, I. J., Strand, S., Smith, P., & Fernandes, C. (2007). Intelligence and educational achievement. Intelligence, 35(1), 13–21.
16. Strenze, T. (2007). Intelligence and socioeconomic success: A meta-analytic review. Intelligence, 35(5), 401–426.
17. Calvin, C. M., Deary, I. J., et al. (2011). Intelligence in youth and all-cause-mortality: Systematic review with meta-analysis. International Journal of Epidemiology, 40(3), 626–644.
18. Trahan, L. H., Stuebing, K. K., Fletcher, J. M., & Hiscock, M. (2014). The Flynn effect: A meta-analysis. Psychological Bulletin, 140(5), 1332–1360.
19. Pietschnig, J., & Voracek, M. (2015). One century of global IQ gains: A formal meta-analysis of the Flynn effect (1909–2013). Perspectives on Psychological Science, 10(3), 282–306.
20. Bratsberg, B., & Rogeberg, O. (2018). Flynn effect and its reversal are both environmentally caused. PNAS, 115(26), 6674–6678.
21. Stanovich, K. E., & West, R. F. (2008). On the relative independence of thinking biases and cognitive ability. Journal of Personality and Social Psychology, 94(4), 672–695.
22. Duckworth, A. L., Quinn, P. D., Lynam, D. R., et al. (2011). Role of test motivation in intelligence testing. PNAS, 108(19), 7716–7720.
23. Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of African Americans. Journal of Personality and Social Psychology, 69(5), 797–811. (For the ongoing debate, see Shewach, Sackett, & Quint, 2019, Journal of Applied Psychology, 104(12), 1514–1534.)
24. Ritchie, S. J., & Tucker-Drob, E. M. (2018). How much does education improve intelligence? A meta-analysis. Psychological Science, 29(8), 1358–1369.
25. Melby-Lervåg, M., & Hulme, C. (2013). Is working memory training effective? A meta-analytic review. Developmental Psychology, 49(2), 270–291.
26. Simons, D. J., Boot, W. R., Charness, N., et al. (2016). Do "brain-training" programs work? Psychological Science in the Public Interest, 17(3), 103–186.
27. Deary, I. J., Whalley, L. J., et al. (2000). The stability of individual differences in mental ability from childhood to old age. Intelligence, 28(1), 49–55.