Monday, August 29, 2022

How real is the Flynn effect?


Changes in mean IQ between 1909 and 2013 (Pietschnig and Voracek 2015, p. 285)



Because of the Flynn effect, average IQ has risen by 35 points over the past century. That’s more than the difference between the threshold of mental retardation and the current average. Does that seem plausible?


In a 1984 paper, James Flynn showed that the mean IQ of White Americans rose by 13.8 points between 1932 and 1978 (Flynn 1984). When that increase, now called the Flynn effect, was charted between 1920 and 2013, the gain in IQ was found to be no less than 35 points (Pietschnig and Voracek 2015).


The IQ gain did not happen at a uniform rate. It can be broken down into five stages:


·         a small increase between 1909 and 1919 (0.80 points/decade)

·         a surge during the 1920s and early 1930s (7.2 points/decade)

·         a slower pace of growth between 1935 and 1947 (2.1 points/decade)

·         a faster one between 1948 and 1976 (3.0 points/decade)

·         a slower pace thereafter (2.3 points/decade)


The Flynn effect began in the core of the Western world and is now ending there. In fact, it has ended altogether in Norway and Sweden and has begun to reverse itself in Denmark and Finland (Pietschnig and Voracek 2015, pp. 283, 288-289).


Was it a real increase?


Average IQ has thus risen by 35 points over the past century. That’s more than the difference between the threshold of mental retardation and the current average. Does that seem plausible?


My mother went to high school during the 1930s, and I went during the 1970s. So my generation should be 13.8 points smarter than hers. That’s a big difference, and it should have been obvious to someone like myself who knew people from both generations.


It wasn’t obvious. My mother had a small library of books that she often consulted, mostly religious literature and works like Welcome Wilderness and Little Dorrit. Not all of her generation were obsessive readers, but many were. And the books they read weren’t light reading. Fiction typically had complex plots with subplots running alongside each other, and religious books were a maze of Biblical references that would seem obscure unless you knew the Bible, usually the King James Version. If you could handle that, you could handle string theory.


The Flynn effect also implies that post-millennials are 10 points smarter than my generation. Again, that’s not my impression. Books and movies now have simpler plots and use a smaller vocabulary—a key component of verbal intelligence. According to the General Social Survey, vocabulary test scores fell by 7.2% between the mid-1970s and the 2010s among non-Hispanic White Americans. The decline affected all levels of educational attainment, so it wasn’t just a matter of dumber people now going to college (Frost 2019; Twenge et al. 2019). The same period also saw an increase in reaction time: since the 1970s, successive birth cohorts have required more time, on average, to process the same information (Madison 2014; Madison et al. 2016). 


Finally, there is the genetic evidence, specifically alleles associated with high educational attainment. In Iceland, those alleles have become steadily fewer in cohorts born since 1910 (Kong et al. 2017). The same trend has been observed between the 1931 and 1953 birth cohorts of European Americans (Beauchamp 2016). According to the Icelandic study, the downward trend is happening partly because more intelligent Icelanders are staying in school longer and postponing reproduction. But it is also happening among those who do not pursue higher education. Modern culture seems to be telling people that children are costly and bothersome, and that message is most convincing to people who like to plan ahead.


Some writers have argued that the genetic decline in intellectual potential has been more than offset by improvements to our learning environment, particularly better and longer education. This improved environment is helping us do more with our intellectual potential. But is there real-world evidence that we are, on average, becoming smarter? Robert Howard (1999, 2001, 2005) cites four lines of evidence:


·        The prevalence of mild mental retardation has fallen in the US population and elsewhere.

·        Chess players are reaching top performance at earlier ages.

·        More journal articles and patents are coming out each year.

·        According to high school teachers who have taught for over 20 years, “most reported perceiving that average general intelligence, ability to do school work, and literacy skills of school children had not risen since 1979 but most believed that children's practical ability had increased” (Howard 2001).


The above evidence is debatable, as Howard himself acknowledges. Fewer children are being diagnosed as mental retarded because that term has become stigmatized. Prenatal screening has also had an impact. As for chess, it’s a niche activity that tells us little about the general population. More journal articles are indeed being published each year, but the reason has more to do with pressure to “publish or perish.” Finally, teachers are not objective observers: they are part of a system that rewards certain views and penalizes others. And if they reject that system, they probably won’t stick around for more than twenty years.


A last word


I suspect we’re getting better at some cognitive tasks, particularly the ones we learn at school—if only because we’re spending more of our lifetime in the classroom. One of those tasks is sitting down at a desk and taking a test. We’re better not only at that specific task but also at the broader one of thinking in terms of questions and answers. Previously, we just learned the rules and imitated those who knew better than us.


Test-taking certainly made an impression on my mental development. Long after my undergrad studies I would have nightmares of sitting alone in an immense exam hall and not knowing the answer to an insoluble question.




Beauchamp, J.P. (2016). Genetic evidence for natural selection in humans in the contemporary United States. Proceedings of the National Academy of Sciences. 113(28): 7774-7779.  


Flynn, J.R. (1984). The mean IQ of Americans: Massive gains 1932–1978. Psychological Bulletin 95(1):29–51.   


Frost, P. (2019). Why is vocabulary shrinking? Evo and Proud, September 11.


Frost, P. (2020). From here it’s all downhill. Evo and Proud, March 16.


Howard, R. W. (1999). Preliminary real-world evidence that average human intelligence really is rising. Intelligence 27: 235–250.  


Howard, R. W. (2001). Searching the real world for signs of rising population intelligence. Personality and Individual Differences 30: 1039–1058.


Howard, R. W. (2005). Objective evidence of rising population ability: A detailed examination of longitudinal chess data. Personality and Individual Differences, 38(2), 347–363.


Kong, A., M.L. Frigge, G. Thorleifsson, H. Stefansson, A.I. Young, F. Zink, G.A. Jonsdottir, A. Okbay, P. Sulem, G. Masson, D.F. Gudbjartsson, A. Helgason, G. Bjornsdottir, U. Thorsteinsdottir, and K. Stefansson. (2017). Selection against variants in the genome associated with educational attainment. Proceedings of the National Academy of Sciences 114(5): E727-E732.


Madison, G. (2014). Increasing simple reaction times demonstrate decreasing genetic intelligence in Scotland and Sweden, London Conference on Intelligence. Psychological comments, April 25 #LCI14 Conference proceedings.    


Madison, G., M.A. Woodley of Menie, and J. Sänger. (2016). Secular Slowing of Auditory Simple Reaction Time in Sweden (1959-1985). Frontiers in Human Neuroscience, August 18.


Pietschnig, J., and M. Voracek. (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.


Twenge, J.M., W.K. Campbell, and R.A. Sherman. (2019). Declines in vocabulary among American adults within levels of educational attainment, 1974-2016. Intelligence 76: 101377.


Santocool said...

I like to compare the Flynn Effect to the development of Artistic Gymnastics, one of my favorite sports.

Some of the least difficult exercises today would have been some of the most difficult in Nadia Comaneci's days.

What does that mean?

That Nadia Comaneci wouldn't be able to compete with the elite of current artistic gymnastics?

Not necessarily.
In fact, you can't say yes or no.
But the idea that the best gymnasts of the 70s would not be able to compete with the best gymnasts of today, just because, at the time, the level of difficulty was much lower, does not seem true.

So if people who took the tests in the first half of the 20th century did the tests today, I highly doubt they would score very low.

The Flynn Effect talks more about the changes that have happened in the tests than with people.

On the basis of fertility patterns alone, among cognitive classes, it seems evident that there has been a decline, accompanied by a demographic transition.

Tom519 said...

I think Charles Murray suggested the cause of the Flynn effect was the percentage of exceptionally low scores has gone way up. The reason being near universal literacy and the experience with taking tests. Those people would have likely scored close to the statistical average with their cohorts if they had basic literacy and experience with written tests. People with average intelligence will score below their ability even on that don't require verbal fluency.
Ashley Montague once claimed that Southern whites scored evenly with Northern blacks on IQ tests. But I haven't been able to find the source for that.

Marco said...

To what extent does the Flynn effect prompt IQ researchers to reevaluate how they are doing their measurements? There seems to be a general recognition that there is a systematic effect that is causing drift and that systematic effect is problematic if researchers want to compare populations across time. Surely, presence of a systematic effect is also problematic if researchers want to compare today’s populations across space. You mentioned that tests of vocabulary and reaction times have been decreasing since the 70’s. Does this mean that vocabulary and reaction times should be incorporated into IQ tests as they measure the underlying quality of interest with greater invariance? If I have heard correctly, different IQ tests emphasize different measurement methods (raven matrices, repeating numbers backwards ...); do all the measurement methods show an equal magnitude for drift over the past century? If some measurement methods showed less drift over the past century, then would that not imply that those measurement methods are measuring the underlying quality of interest with greater fidelity? We have very good reason to believe our ancestors were not mentally retarded. We may have moderately good reason to believe genetic data on intelligence. If the current IQ “meter” does not reflect what we know to be true, we can ask what is affecting the meter but I would also be very curious about what can be done to the meter to make it less susceptible to drift. The proper measurement goal would be to have all data from all sources agree and so have confidence in identifying real intelligence differences across time and space.

Peter Frost said...


What about performance-enhancing drugs?


He was referring to the U.S. Army Alpha and Beta tests (1919):

Alper, T. G., & Boring, E. G. (1944). Intelligence test scores of Northern and Southern white and Negro recruits in 1918. The Journal of Abnormal and Social Psychology, 39(4), 471–474.


Klineberg (see 9: 4717) and recently Benedict and Weltfish (see 18: 787) selected extreme states to show that intelligence test scores are more a function of geography than of skin color. The complete picture is presented in this note, with an analysis of variance of the Army Alpha and Beta tests in all states in 1919. Inclusion of the Beta results helps the whites more than the Negroes. The complete results show test scores to be significantly related both to skin color and to locale.


It's difficult to compare the results of IQ tests that have been conducted under different conditions, in different cultural contexts, and at different points of time. In general, I consider IQ data to be suggestive rather than conclusive. An exception would be controlled adoption studies, like the one done in Minnesota. Future research should focus on polygenic data (alleles associated with educational attainment). Even that kind of data can be problematic because the alleles have been identified in European-descended subjects.

Marco said...

Thanks for your reply. My interpretation is that you are suggesting that the technology of IQ tests is best interpreted along the lines of – given a time and environment, a test score further above the mean is suggestive of higher general cognitive ability and one further below the mean is suggestive of lower general cognitive ability. Seeking stable measurements will probably not be a productive avenue for future effort, but the existing relative scores can still provide useful information for practical purposes, (such as identifying talent).
Then you seem to suggest that you believe, in the future, more insightful measurements of cognitive ability will be extractable from genome data and advancements in understanding should be focused there.
If I am not misinterpreting you views, then that is fair enough. I may be asking too much from the technology of IQ testing. In fact, there is nothing wrong with a relative indicator of any quantity, even in the hard sciences. Sometimes direct quantities of interest are impossible to measure and one measures something related to but not quite the same as what one would like to. Any measurement is useful to the extent it can direct action in the real world.

Santocool said...


What about performance-enhancing drugs?''

I think these days are not allowed. That's why female gymnasts are so muscular these days, with the exception of Chinese gymnasts [different choices of body types and skills]. They need to be very strong to do so much acrobatic exercises.

TomR said...

Intelligent people have moved from farms and small cities, where they were doing jobs like farming or crafts, to cities where they were given like engeneering, business, science jobs. A better usage of existing population, including better mental training.
Then there was this change in mating (at the same time as above) from based on geographical proximity (same or neighboring village) to based on same education class. This changed the new generation into more spread distribution.
There's also popularity of abortion that decreased numbers of genetically unhealthy offspring.
Brain-drain type migrations may also help first-world countries in increasing their overall number of intelligent people even if native population trends were dysgenic.
In case of Eastern Europe etc. it was also first time how it is when people are not killed in frequent wars or persecutions.

Pumpkin Person said...

People wouldn't wonder so much why IQ scores are going up if they realized how little they have: