Surprising evidence on the Flynn Effect

1. Non-verbal IQ has risen more rapidly than has verbal IQ.

2. Performance gains are smallest on the most culturally specific tests, and largest on the most abstract tests.

3. Performance gains, as they occur over time, are roughly constant for all age groups.

4. Problem-solving abilities have seen the biggest performance gains.

5. Gains on the "Ravens" test started occurring before the TV era, much less the computer game era.

#3 is perhaps the biggest surprise to me, as it contradicts most of the obvious explanations for the Flynn effect. 

Those results are summed up in the very interesting "The Flynn effect and its relevance to Neuropsychology," by Merrill Hiscock, Journal of Clinical and Experimental Neuropsychology, 2007.  Here is Andrew Gelman’s post on that paper.

Hiscock puts it well: "..the Flynn effect constitutes a compelling example of large between-group IQ differences [across generations] that are completely environmental."


Recent papers (Teasdale and Owen, Sundet Barlaugb and Torjussen )have suggested the Flynn effect stopped around 1990 in Northern Europe, though this could be related to immigration by groups that test lower on IQ tests.

Maybe it's just me, but I've never understood "verbal IQ" as a measure of intelligence. So I haven't been
exposed to that particular word before, that means I'm less intelligent? Less *knowledgeable*, sure,
but why less intelligent?

1. Non-verbal IQ has risen more rapidly than has verbal IQ.

This isn't surprising at all. Are novels of the 21st century appreciably better than those of the 19th century? It's arguable.

But the leaps in technology—which is done in the currency of math and science—are undeniable.

Or, more succinctly, compare the progress in movie scripts versus those in special effects.

The Flynn Effect does not appear to have reduced differences (in SD units or, hardly, in points) between mean IQ's of different ethnic groups. So it does not suggest that the genetic contribution to IQ has changed. As many scholars have noted, when you minimize variations in environment which contribute to differences in IQ (or anything else), then any differences caused by genetic factors become more, not less visible. What this means for politics I leave to the reader.

[1] Except perhaps on the basis of selection over multiple generations. Future biotech innovations may alter this.

Verbal IQ should not be interpreted literary. It’s also a measure of broad analytical/judgment/creative, that is *measures* using verbal tests (antonyms). This requires you to understand and abstractly think about what words mean.

Of course it also measures language comprehension and word knowledge. What economists often call “intuition† is probably largely verbal IQ.

Think lawyer or public intellectual, rather than linguist or poet (the two other also have high verbal IQ, but the two former show why it matters).

Tino, I understand that, but in order to get a verbal IQ test question right, you generally have to BOTH
be able to reason properly, AND have the requisite language *knowledge*. So missing it means you either
couldn't reason, or didn't know of that word. Since there's such a big reason to miss it for non-intelligence
reasons, I don't see why it's considered an intelligence test.

The test for verbal IQ is basically an advanced vocabulary test.

A person's verbal IQ is largely determined by the number of words and the concepts underlying those words he has learned. But learning words and concepts is a "linear additive set" which means that learning one word/concept gives little advantage in learning the next word/concept. Thus, the learning of vocabulary is not amenable to acceleration through improved teaching techniques. Learning vocabulary is a slow slog that is limited by the person's verbal IQ. People with low verbal IQs learn more slowly than people with high verbal IQs and there is litle we can do (at least currently) to help them catch up.

See Wes Becker's 1977 article from the Harvard Educational Review Teaching Reading and Language to the Disadvantaged—What We Have Learned from Research:

In contrast to the general-case learning involved in decoding, arithmetic, grammar, and spelling-by-sounds, the learning of vocabulary and concepts usually involves a “linear-additive set† (Becker & Engelmann, 1976, p. 58). In a linear additive set, the learning of one element gives little advantage in learning a new element. To be sure, there are families of words that have common root meanings and common meanings of affixes, which permit some limited general cases to be generated. But, by and large, the learning of proper names, new concepts, and the learning of synonyms for concepts already known by another name, involve linear additive sets in which each new element must be taught. Knowledge of the English language, which is absolutely essential to oral and written comprehension, serves largely to define intelligent behavior (Miner, 1957). Teaching this language involves a task of the first magnitude.

Schools have never had programs in reading that systematically build vocabulary concept knowledge. Except for technical vocabularies, this task has been largely left to parents. Furthermore, since the achievement tests are built by procedures that measure age changes, and not simply the effects of school instruction, children from homes with weak support for language development fall progressively behind on current reading tests. This finding is commonly reported.

Basically verbal, IQ is the limiting factor for how much a person is capable of learning since it determines the speed at which new words/concepts are learned which determines the person's ability to comprehend the increasingly sophisticated language in higher learning.

Given what we know about the Flynn it seem safe to assume that our ancestors group IQ was lower than that Americans and Europeans today and probably below Tyler's villagers. However they achieved a lot economically and invented a lot of useful new technologies, although they didn't do much academically and got into a lot of drunken accidents.

The Flynn effect implies that mean IQ 100 years ago was much lower than it is today; however, there are plenty of examples of genius form those times that we can draw on. So, assuming *mean* IQ has been rising, does that mean the entire distribution has been shifted higher, or merely that its shape has changed? We know that *mean* IQ is higher, but does that necessarily imply that we have more geniuses than before? In other words, how normal is the distribution - what has been happening in the tails? The really useful information regarding the Flynn effect would be more detailed information on how the IQ distribution within ethnic groups has changed over time. We know that the *mean* has been changing - what about the *variance*, the *skew*, and the *kurtosis*? Assuming the Flynn effect is due to better nutrition, this info could be of extreme usefulness in predicting how the average IQ of currently malnourished countries will change if nutrition improves.

P.S. for a graphical illustration of mean, variance, skew, and kurtosis, check here:

Right, as a topic for economists looking for something important and powerful to research, IQ is the twenty-dollar bill lying on the sidewalk while the entire economics profession walks by.

Some economist should write about this as a classic example of market failure.

"Which tool, however, will continue to be underexploited in the field of economics for careerist reasons? You guessed it: IQ."

Can't you comment without speculating on people's motives? All this post does is summarize some facts about the Flynn Effect and you can't resist turning the comment box into a speculations about others motives.

Stephen B: of course he could. But why should he?

The motivations here are, after all, so very interesting.

"I understand that, but in order to get a verbal IQ test question right, you generally have to BOTH
be able to reason properly, AND have the requisite language *knowledge*. "

Mwah, doesn't the same hold for quantitative tests? If you haven't encountered the square root sign or Pygatoras' theorem before, you won't solve many geometry questions.

--But the leaps in technology—which is done in the currency of math and science—are undeniable.

this means what? Technology gets to build on the shoulders of giants, doesn't require MORE IQ to keep moving forward at a faster rate. let's look at the 17th and 18th centuries. For a group of people supposedly a lot less smart, their sophistication with ideas was no less than ours now, either by average or top quintiles.

Dear iam,

I hate to break this to you, but the real world doesn't look at all like your understanding of economic theory says it should look like: low IQ countries haven't gotten very big gains from trade, as their low per capita GDPs show. In contrast, high IQ countries with Northeast Asian populations have gotten big gains from trade, and have consistently pushed into higher technology industries.

As I've said, when it comes to research topics, IQ is the $20 bill lying on the sidewalk while the whole economics profession walks by with their noses in the air, too refined to deign to notice it.

Comments for this post are closed