RSS
I haven’t seen confirmation of this yet, but it appears that the philosopher-scientist James Flynn, discoverer of the Flynn Effect of rising raw IQ scores, has died.
Flynn was not the first to notice that IQ test creators had to periodically make their tests harder to keep their average score at 100, but, like Columbus, after him the Flynn Effect stayed discovered.
Numerous leftists complain at length about the science of intelligence, but Flynn was the only man-of-the-left of the latter 20th Century to make a major empirical discovery in intelligence research, one that rightly lowered the confidence of psychometricians.
The lengthy epistolary dialog between James Flynn and Arthur Jensen over Flynn’s discovery of rising raw IQ scores should be taught in Philosophy of Science courses as an example of how to do science right. Flynn brought the Flynn Effect (as it was named in The Bell Curve) to the attention of Arthur Jensen in the 1970s (IIRC). Jensen was initially skeptical about the significance of Flynn’s findings, and offered a list of scientific challenges. Over a number of years of research, Flynn patiently answered each one, earning Jensen’s admiration.
I’ve written a couple of times at length about the Flynn Effect
- Here’s my 2007 book review in VDARE.com of Flynn’s What Is Intelligence: Beyond the Flynn Effect
- Here’s my 2015 Unz Review post on “The Flynn Effect: IQ Testing Across Space and Time“
In short, tremendous efforts had been made to make IQ testing more valid across space (e.g., to avoid cultural bias, the non-verbal Raven’s Progressive Matrices invented in the 1930s look like something that would be inscribed on the monolith from 2001). But little attention had been devoted to keeping the scores of IQ tests stable across time, with the Raven’s being particularly susceptible to a huge Flynn Effect, with raw IQ scores rising several points per decade in the 20th Century.
[Repost now that comments are enabled here.]
The Raven’s vs. vocab/info graph nicely demonstrates what has been obvious to me, but which doesn’t seem to register with a lot of people when discussing the Flynn effect:
There is no significant “Flynn effect” rise in actual “intelligence”.
I think there were some real gains in actual intelligence in early/mid-20th century from better nutrition coupled with improved education and more information (radio, TV) prompting denser wiring. (More or less akin to how Americans got taller in the same time period.) But …
The Flynn effect is mostly measuring increased exposure to “symbolic processing”.
Basically people are doing wildly better on Raven’s like tests because they have learned how to do Raven’s like tests. And Raven’s type tests are super-subject to learning.
This is unfortunate, because Raven’s type tests are an excellent way to get around language issues and test IQ “cross-culturally”. (And i do really well on them–making me “smart”.) They would be really good tests if no one taking them had ever been exposed to such “symbolic processing”. But the results are really noisy–actually just mediocre. Because the level exposure to symbolic processing, and in fact Raven’s type puzzles, varies so wildly across nations, and, as we see, across time.
~
The simplest way to get a handle on cross-national, cross-racial IQ is simply to look at the societies that various people produce. There isn’t an “information flow” problem anymore. No much confusion–well outside the West!–about what’s required to make your nation a decent place, nor the technology for material prosperity. What you see from a nation gives a pretty decent picture of what its people are capable of. At least to first order: If it sucks … the people suck.
But if you want a good IQ proxy, i’d say the PISA tests–particularly the math part. Math is cross cultural. Every society is trying to give their children a math education So with the caveats of nutrition and educational quality, the math PISA is a pretty good proxy for “how smart are those folks”.
Finally, the most fair and direct racial/ethnic IQ scores are simply looking at them within a 1st world nation like the US, where everyone’s well fed and gets a similar basic school program. The issues of admixture and selective immigration are there. But basically what you see here is what is. If you can’t make it here … you can’t make it anywhere.
~~
The basic thing to keep in mind:
We have no direct measure of intelligence–like we do of say height. The way we measure intelligence is to give people mental tasks–IQ tests. But IQ test tasks–like any mental tasks–can be learned. People get better with practice. And on some tasks–like Raven’s–people get wildly better with more exposure and practice.
Maybe someday we’ll have something direct. You’ll roll into the interview center and then plug some electrodes onto you–like they can do in the auto shops to see what the engine computer is seeing–load your brain a bit and see what’s under the hood and if you’ve got what it takes:
“Sorry AnotherDad”, you’re neurons are misfiring a lot. We’ve got your old results and frankly … you ain’t 22 anymore! Stay retired.”
US public school amumni are bad at foreign languages compared to alumni in Germany, the Netherlands , Poland or Scandinavia. That's not because they are stupid, it's because they have ridiculous foreign language curricula that simply don't teach you much and don't expect you to really be fluent in the language. German Gymnasium students all lear at least 2 foreign languages (and some learn three).
Foreign language learning has a high g-loading, as does math, but only when you compare like with like, e.g. people who were exposed to the same school curricula.
I wish to second the comment by AnotherDad. IQ scores have risen over the past century because people have become more familiar with test-taking.
When I did fieldwork in the late 1980s among elderly French Canadians, I was warned against asking straight questions and expecting to get straight answers. “In the past, people just didn’t think that way.” This was partly because straight questions were often seen as impolite and partly because people generally learned through observation and imitation.
The Flynn effect has actually masked a decline in intelligence. If we look at alleles associated with high educational performance, we find they have actually become less frequent in the population over the past century. This is the conclusion of two “polygenic” studies from Iceland and the United States. It is also the conclusion of several studies on reaction time.
We’ve compensated for this decrease in ability through an increase in education: “the jockey has become more skilled as the power of the horse dwindles.” Unfortunately, this work-around has its limits. Indeed, the Flynn effect is levelling off in much of the Western world, perhaps because we have run out of educational wiggle room.
Reference
https://evoandproud.blogspot.com/2020/03/from-here-its-all-downhill.html
doesn't seem very convincing. that might make sense the first few decades, and i'm somewhat onboard with that idea. it wouldn't explain things long term though.
how much MORE familiar with standardized testing could the students be at this point? they've had it rammed down their throats for 50 years. they're as familiar with it as possible. yet the numbers slowly went up decade by decade anyway.
it's like saying not many people are familiar with bench press, and that's why the strength sports numbers keep going up. like, who is not familiar with the basic lifts at this point? yet the numbers slowly keep getting higher decade by decade.
the smartest people are, probably, still getting slightly smarter. it's also possible that at the same time, dumb people are having more dumb kids than everybody else. both things can be occurring at the same time. and likely are. the width of the bottom is expanding, but the ceiling of the top is also rising.
the guys who make the 'technology has stalled' argument, i'm pretty sure they just don't know what they're talking about. not only do i subscribe pretty heavily to the same idea that Anatoly Karlin does about increasing difficulty, some of the next difficulty level stuff is already in the process of happening. if they think 2070 will look 'about the same' as 1970, they're just dead wrong. 2020 already looks A LOT different than 1970.
i’m under the impression that what motivated Flynn was being Irish, and wondering why the Irish were so far behind British people and other europeans in general. he did a good faith, decades long investigation into topics like this, which seemed to be a recurring theme in his career. later on he got into trying to understand africans in America.
he always seemed to root for the underdog, but he never fudged his work to produce inaccurate results. he was trying to see if people way behind could catch up, or even catch up all the way. a lot of the time they got better, but mostly they never caught up all the way.
“IQ scores have risen over the past century because people have become more familiar with test-taking.”
doesn’t seem very convincing. that might make sense the first few decades, and i’m somewhat onboard with that idea. it wouldn’t explain things long term though.
how much MORE familiar with standardized testing could the students be at this point? they’ve had it rammed down their throats for 50 years. they’re as familiar with it as possible. yet the numbers slowly went up decade by decade anyway.
it’s like saying not many people are familiar with bench press, and that’s why the strength sports numbers keep going up. like, who is not familiar with the basic lifts at this point? yet the numbers slowly keep getting higher decade by decade.
the smartest people are, probably, still getting slightly smarter. it’s also possible that at the same time, dumb people are having more dumb kids than everybody else. both things can be occurring at the same time. and likely are. the width of the bottom is expanding, but the ceiling of the top is also rising.
the guys who make the ‘technology has stalled’ argument, i’m pretty sure they just don’t know what they’re talking about. not only do i subscribe pretty heavily to the same idea that Anatoly Karlin does about increasing difficulty, some of the next difficulty level stuff is already in the process of happening. if they think 2070 will look ‘about the same’ as 1970, they’re just dead wrong. 2020 already looks A LOT different than 1970.
how much MORE familiar with standardized testing could the students be at this point?
Uh, the Flynn effect is levelling off throughout most of the Western world. So greater familiarity is not producing any pay-off. We’ve reached the glass ceiling. We could double the number of years people spend at university, and there would be no improvement at all.
A second possibility is that the ceiling itself is getting lower. Greater familiarity with test-taking is perhaps being negated by a decline in average cognitive ability.
In any case, the effect of greater familiarity is not a real increase in intelligence.
RIP.
Flynn to be played by … Sterling Hayden!
The Raven’s vs. vocab/info graph nicely demonstrates what has been obvious to me, but which doesn’t seem to register with a lot of people when discussing the Flynn effect:
There is no significant “Flynn effect” rise in actual “intelligence”.
I think there were some real gains in actual intelligence in early/mid-20th century from better nutrition coupled with improved education and more information (radio, TV) prompting denser wiring. (More or less akin to how Americans got taller in the same time period.) But …
The Flynn effect is mostly measuring increased exposure to “symbolic processing”.
Basically people are doing wildly better on Raven’s like tests because they have learned how to do Raven’s like tests. And Raven’s type tests are super-subject to learning.
This is unfortunate, because Raven’s type tests are an excellent way to get around language issues and test IQ “cross-culturally”. (And i do really well on them–making me “smart”.) They would be really good tests if no one taking them had ever been exposed to such “symbolic processing”. But the results are really noisy–actually just mediocre. Because the level exposure to symbolic processing, and in fact Raven’s type puzzles, varies so wildly across nations, and, as we see, across time.
~
The simplest way to get a handle on cross-national, cross-racial IQ is simply to look at the societies that various people produce. There isn’t an “information flow” problem anymore. No much confusion–well outside the West!–about what’s required to make your nation a decent place, nor the technology for material prosperity. What you see from a nation gives a pretty decent picture of what its people are capable of. At least to first order: If it sucks … the people suck.
But if you want a good IQ proxy, i’d say the PISA tests–particularly the math part. Math is cross cultural. Every society is trying to give their children a math education So with the caveats of nutrition and educational quality, the math PISA is a pretty good proxy for “how smart are those folks”.
Finally, the most fair and direct racial/ethnic IQ scores are simply looking at them within a 1st world nation like the US, where everyone’s well fed and gets a similar basic school program. The issues of admixture and selective immigration are there. But basically what you see here is what is. If you can’t make it here … you can’t make it anywhere.
~~
The basic thing to keep in mind:
We have no direct measure of intelligence–like we do of say height. The way we measure intelligence is to give people mental tasks–IQ tests. But IQ test tasks–like any mental tasks–can be learned. People get better with practice. And on some tasks–like Raven’s–people get wildly better with more exposure and practice.
Maybe someday we’ll have something direct. You’ll roll into the interview center and then plug some electrodes onto you–like they can do in the auto shops to see what the engine computer is seeing–load your brain a bit and see what’s under the hood and if you’ve got what it takes:
“Sorry AnotherDad”, you’re neurons are misfiring a lot. We’ve got your old results and frankly … you ain’t 22 anymore! Stay retired.”Replies: @Occasional lurker
‘Every society is trying to give their children a math education So with the caveats of nutrition and educational quality, the math PISA is a pretty good proxy for “how smart are those folks”.’
No, it’s worse than Raven’s matrices, if what you are interested in is some sort of innate ability and not something learned. Why? Different countries have very different math curricula. The correlation between strenuous math curricula and coutrywide performance in the easy PISA type math must be around 0.9. Same with TMSS.
US public school amumni are bad at foreign languages compared to alumni in Germany, the Netherlands , Poland or Scandinavia. That’s not because they are stupid, it’s because they have ridiculous foreign language curricula that simply don’t teach you much and don’t expect you to really be fluent in the language. German Gymnasium students all lear at least 2 foreign languages (and some learn three).
Foreign language learning has a high g-loading, as does math, but only when you compare like with like, e.g. people who were exposed to the same school curricula.
THE JOLLY HERETIC CELEBRATES THE LIFE OF THE LATE HONEST LIBERAL, JAMES FLYNN