Guide
The Flynn effect
For most of the 20th century, people got better at IQ tests. Not slightly better: measured averages climbed by around three points a decade. This is the Flynn effect. It is one of the most important findings in the history of testing, and it explains why every IQ test has to be re-standardized from time to time.
Want your own number? The test is 25 visual puzzles, about 12 minutes, no sign-up.
What was found
IQ tests are scored against a reference sample: a large group of people tested when the test was built, standardized so their average is 100 (Wikipedia: Intelligence quotient). When researchers gave old tests to new generations, or compared scores across editions, they found something surprising. Later generations scored higher against the old reference groups, and the gap kept growing.
In 1984, James Flynn published a study in Psychological Bulletin documenting gains of about 3 IQ points per decade in the United States. Those gains have since been studied across the world. A 2015 meta-analysis pooled 271 samples, about 4 million people from 31 countries, tested between 1909 and 2013. It estimated an average gain of about 2.8 points per decade and found that the gains have slowed in recent decades (Pietschnig & Voracek, 2015).
A reversal in some places
The rise has not continued everywhere. In Norwegian military conscript data, the trend reverses for cohorts born after about 1975: scores start to decline rather than rise (Bratsberg & Rogeberg, 2018). Researchers call this the negative Flynn effect.
One country's data does not describe the world, and the overall picture from the 2015 meta-analysis is of slowing gains rather than a universal decline. What both findings make clear is that average test performance is not fixed. It moves over decades.
Why scores rose, and whether they still do
The gains were not even across abilities. The 2015 meta-analysis estimated annual gains of 0.41 IQ points for fluid reasoning, 0.30 for spatial, 0.28 for full-scale and 0.21 for crystallized performance (Pietschnig & Voracek, 2015). Its authors concluded that factors linked to life history speed mainly explain the overall trajectory, with social multiplier effects and economic prosperity explaining differences between domains (Pietschnig & Voracek, 2015).
Not everyone sees a slowdown. A 2014 meta-analysis of 285 studies estimated 2.93 points per decade on modern Stanford-Binet and Wechsler tests and called that inconsistent with the idea that the effect is fading (Trahan et al., 2014). Where declines have appeared, as in Norway, a study of brothers found the rise and the decline within families, pointing to environmental causes (Bratsberg & Rogeberg, 2018).
Try one · Abstract Reasoning
Which tile completes the grid?
Tap the option you think is right, or show the answer.
Show the answer
The missing tile has a large outlined square (option 6).
How to solve it
- Start with the first column and name what changes from tile to tile.
- Every row and every column contains each shape (triangle, square and diamond) exactly once.
- Every row and every column contains each fill (outline, hatched and solid) exactly once.
- The size (large) stays the same in every tile.
- Check the rules on the second column, then apply them to the third.
- So the missing tile has a large outlined square.
That was 1 puzzle. The full test has 25.
Why it matters for every IQ score
Remember what an IQ score is: your position compared with a reference sample. If the population's average performance rises after that sample was tested, today's typical person outperforms the old average. A test with old norms will then give average people scores above 100.
A rough illustration: at about 3 points per decade, a test whose reference sample was collected 20 years earlier would read about 6 points high for a typical person taking it today. That is larger than the measurement error of a good clinical test.
The fix is re-standardization: publishers collect new reference samples for new editions so that 100 keeps meaning average. The Wechsler adult scale, for example, has been revised repeatedly, with WAIS-IV published in 2008 and WAIS-5 in 2024 (Wikipedia: WAIS). When you compare an old score with a new one, the edition and its norms matter.
Comparing scores across decades
If you are comparing a score from years ago with a recent one, or your score with a parent's, the Flynn effect is one of several reasons the numbers may not line up:
- Norm dates. Find out which edition was used and roughly when it was normed. Older norms tend to give higher numbers for the same performance during periods when averages were rising.
- Scales. Older Stanford-Binet forms used a standard deviation of 16 and the Cattell Culture Fair III scale uses 24 (Wikipedia: IQ classification), so convert to percentiles before comparing.
- Measurement error. Clinical scores are reported with a confidence interval, often around plus or minus 5 points at 95% (Wechsler, 2008). Small gaps between two scores are usually noise.
None of these adjustments is exact, so treat cross-decade comparisons as approximate. The percentile to IQ converter handles the scale part.
What the Flynn effect does not tell you
- It is about averages, not individuals. It describes how whole cohorts performed. It tells you nothing about any one person's score.
- It does not settle what caused it. Researchers have proposed many explanations, and the debate is ongoing. We only cite what the studies above measured.
- It does not mean a score is meaningless. Within a given test and its norms, scores still rank people consistently. The Flynn effect is a reason to keep norms current, not a reason to ignore scores.
What it says about tests themselves
Gains of nearly three points a decade, sustained over several generations, show that average test performance can change a great deal. That is a useful reminder whenever someone treats an IQ score as a fixed quantity. Our guide to what an IQ test measures covers what scores can and cannot tell you, and culture-fair tests discusses how test design tries to reduce the influence of background.
How IQ Octopus handles it
IQ Octopus reports on the standard scale, mean 100 and standard deviation 15. Our puzzle difficulty values start from design settings and are refined from accumulated response data over time, so the comparison reflects people taking the test now rather than a reference group frozen decades ago. Read the full method on how it works, or see how averages are defined in what is the average IQ.
Sources
- Pietschnig, J., & Voracek, M. (2015). One century of global IQ gains. Perspectives on Psychological Science, 10(3), 282-306. doi.org/10.1177/1745691615577701
- Bratsberg, B., & Rogeberg, O. (2018). Proceedings of the National Academy of Sciences, 115(26), 6674-6678. doi.org/10.1073/pnas.1718793115
- Intelligence quotient. Wikipedia. en.wikipedia.org/wiki/Intelligence_quotient
- Wechsler Adult Intelligence Scale. Wikipedia. en.wikipedia.org/wiki/Wechsler_Adult_Intelligence_Scale
- IQ classification. Wikipedia. en.wikipedia.org/wiki/IQ_classification
- Wechsler, D. (2008). WAIS-IV Technical and Interpretive Manual. Pearson.
- Trahan, L. H., Stuebing, K. K., Fletcher, J. M., & Hiscock, M. (2014). The Flynn effect: A meta-analysis. Psychological Bulletin, 140(5), 1332-1360. doi.org/10.1037/a0037173