May 15, 2013

Flynn Effect: The smart get smarter

The Flynn Effect is a term coined by Charles Murray and Richard Herrnstein to refer to the phenomena of rising raw scores on IQ tests that James Flynn has frequently documented.

It is widely assumed that the Flynn Effect will narrow racial/ethnic gaps in IQ. The handful of  semi-sophisticated denouncers of Jason Richwine like to cite the Flynn Effect: Well, sure, Hispanics lag in IQ now, but the Flynn Effect will solve that Real Soon Now.

This is not an unreasonable presumption on a priori diminishing marginal returns grounds. For example, in recent decades, life expectancy gaps between countries have mostly narrowed (except during the worst AIDS years in Africa), because traditionally rich countries have had diminishing returns at improving life expectancies compared to traditionally poor countries.

But, there's not much evidence of IQ racial/ethnic gaps closing in the real world: gaps in IQ are notoriously intractable. So, just eyeballing the data over the decades suggests that the Flynn Effect is about the same size for all racial/ethnic groups, with possibly Asians having a larger one.

Here's a major study of the Flynn Effect focusing directly on that question, using an amazing data resource the Children of the National Longitudinal Study of Youth 1979. The females who joined the nationally representative NLSY79 to have their lives tracked by the Bureau of Labor Statistics have had over 15,000 children, and a sizable number are having their children tracked, including IQ testing.

The researchers found sizable Flynn Effects among these children from the mid-1980s to the mid-2000s. (I believe they think they've found a way to adjust for those women taking longer to reproduce tending to be smarter.) But, did they find larger Flynn Effects among blacks and Hispanics than among whites? (The small number of Asians and Native Americans are lumped in with whites in this study.)
Intelligence. 2010 July 1; 38(4): 367–384. 
The Flynn Effect within Subgroups in the U.S.: Gender, Race, Income, Education, and Urbanization Differences in the NLSY-Children Data 
SiewChing Ang, Joseph Lee Rodgers, and Linda Wänström 
Although the Flynn Effect has been studied widely across cultural, geographic, and intellectual domains, and many explanatory theories have been proposed, little past research attention has been paid to subgroup differences. ... These prior findings suggest that the NLSYC data can be used as a natural laboratory to study more subtle FE patterns within various demographic subgroups. We test for subgroup Flynn Effect differences by gender, race/ethnicity, maternal education, household income, and urbanization. No subgroups differences emerged for three demographic categories. However, children with more educated (especially college educated) mothers and/or children born into higher income households had an accelerated Flynn effect in their PIAT-M scores compared to cohort peers with lower educated mothers or lower income households. We interpret both the positive and the null findings in relation to previous theoretical explanations.

... Until the current study, this finding could be explained by a differential Flynn Effect in which minority scores increased at a steeper rate. However, we found no interaction in our data; the three different race categories each showed substantial FE’s, but they also tracked closely to the same consistent increase. The absence of race differences in FE patterns also has implications for the various other theories. If FE patterns in the NLSY-Children emerged from within the family, or were related to average family size (e.g., Sundet, Borren, & Tambs, 2008), ethnic differences in family culture and family size could potentially create differential FE patterns; but those differences were not observed. If average educational quality is lower for minorities, this could lead to differential FE patterns; again, this finding did not obtain. As for gender, theories that are silent with regards race differences in FE patterns are consistent with the current findings, including the nutrition hypothesis, testing artifacts, and heterosis. ...

In conclusion, we now know new and important features of the Flynn Effect, at least the FE found in U.S. data from the past two decades or so. The effect itself is strong and consistent, but we found no differential gender or race FE, nor was there much of a differential urbanization status identified. The positive finding of a differential FE in relation to maternal education (and at a smaller level, household income) at the older ages is suggestive of some of the dynamics of the process leading to the Flynn Effect.

Thus, across this spectacular, nationally representative two-generation database, blacks and Hispanics aren't closing The Gap over a 35 year period

The main subgroup effect seen is that mothers with more education have children with larger Flynn Effects: the smart get smarter, even relatively speaking.


Anonymous said...

Jan te Nijenhuis & Henk van der Flier's meta-analysis of the Flynn effect indicates that gains are negatively correlated with the g factor, while group differences are positively correlated with the g factor. So group differences and the secular rise in IQ likely have different causes. This is consistent with the hereditarian hypothesis for group differences:


Anonymous said...

bio-engineering will do it.

Anonymous said...

Who needs bio-engineering? A voluntary eugenics program would raise NAM IQs in a generation. Simply pay all NAM teens $10 to get implanted with Norplant. When the Norplant wheres off in 5 years, pay the ones who scored in the top 10% on the SAT to have kids with the top 10% NAM men. Pay the rest of the NAM women another $10k to re-up their Norplant for another 5 years.

x said...

flynns effect seems like a big wild goose chase to me. the gains are hollow and i believe measurement invariance does not hold between age cohorts? the gains don't represent genuine rise in intelligence but an artifact of some kind.

Simon in London said...

I remember going through Lynn's IQ data by race & nationality, and it looked as if Europeans outside of far-south-east Europe were overperforming where they 'should' be based on brain size, other Caucasians, etc, by up to 10 points. So it seemed plausible that there were some cultural reasons for this overperformance and other races might close the gap to that sort of extent, though not eliminate it.

There seems little or no evidence of this happening at the low (sub-IQ 100) side, though (it may be that eg African-American IQ 85 is already environment-optimised, compared to eg African-African median IQ 68-70).
However, European IQ overperformance also means that we have a smaller gap with north-east Asians than might be expected from brain size comparison; 5 points rather than 10-15 points. Equalise environmental effects and the implication is that NE Asian median IQ may rise from 105 to the ca 110-115 range, similar to US Ashkenazi. US academic performance indicators may show something like this happening?

Anonymous said...

On a tangential note, Richard Lynn feels that the phenomenon of the secular rise in IQ should be renamed "the Runquist effect" in honor of it's true discoverer E. A. Runquist.


DR said...

Anyone know the breakdown of the Flynn effect for Ashkenazis?

It seems like between 1800 and 2000 the very cognitively demanding fields of human civilization like physics and theoretical math went from relatively small Jewish participation to a very disproportionately high amount.

Either the Flynn effect helped enhanced Ashkenazi intelligence to a greater degree, bringing more overall Jewish brainpower. Or for cultural, institutional, or other reasons Jewish brainpower was there before but wasn't brought to bear on technical endeavors. Maybe prior to the modern era the smartest Jews were directed to become rabbis instead of mathematicians.

Anonymous said...

Or for cultural, institutional, or other reasons Jewish brainpower was there before but wasn't brought to bear on technical endeavors. Maybe prior to the modern era the smartest Jews were directed to become rabbis instead of mathematicians.

That's the reason. Jews were not really involved with European high culture until the 19th century. Jewish achievement in science and culture is about their assimilation to European culture and taking up European pursuits such as science and the arts.

Anonymous said...

I think that it's hilarious that the Flynn effect is touted as some sort of anti-IQ trump card by liberals. Does it even exist? Why was the 1996 renorming of the SAT downward and not upward if people are getting smarter all the time? And, even if it does, why does the IQ gap between the races over time seem to be as constant as the acceleration of gravity or the value of pi?

sunbeam said...

I've heard of the Flynn effect. It has been discussed in many places, and has enough data from IQ tests to have a debate about it for a number of years.

But my question is, where exactly is the Flynn effect happening?

I see no evidence of the Flynn effect in my daily life. In fact it is exactly the opposite.

I can't think of anything other than use a computer that the younger generations do better than the ones that preceded them.

It definitely isn't math, or at least arithmetic. It isn't geography (Ever notice how many kids have no sense of direction? How easily they get lost? Can't find a place they've been to a dozen times if they have to go there alone?).

Manual skills are on the decline. I have real doubts about the ability of modern kids to change a tire, let alone do the kind of scratch carpentry that was as accepted as breathing in past years. And they have no capacity to figure things out either. I mean give them a bunch of pvc pipe and fitting and tell them to go from A to B. No telling what you will get.

It's not reading. For the most part they don't read. It's not general knowledge or history. They couldn't write a short story or express an argument in written form to save their lives.

I live in the South. Some of this has always been this way, but now it is worse. And some of the things they were able to do, they can't seem to do anymore.

Obviously the young people I encounter aren't in the HYPS pipeline.

But shouldn't I see something somewhere? It sure looks to me like people are getting dumber not smarter.

Chicago said...

Has the gap between Europeans and native aborigines really changed at all in the hundreds of years since the first contact was made? There's been the rise of hybrid populations which muddy the waters but overall it seems to have been pretty consistent throughout.

pat said...

I understand that Lynn now calculates that the Flynn effect is only on phenotypic IQ and that because of various dysgenic effects that genotypic IQ has been dropping at the rate of .43 points per generation.

I must admit I don't quite understand what all this means. If it means anything at all.


David said...

Maybe some rising tide is lifting all boats (tho I don't see this in life), but the dingy still floats below the yacht.

Forgive the bad question, but is the Flynn Effect possibly an artifact of the increasing social stratification Murray and Herrnstein referred to in TBC? Does clustering smarts with smarts (and dumbs with dumbs) have a bigger effect on their individual and group performance than clustering dumbs with smarts has?

It seems to me that I step a little livelier - I'm a bit more on my game - when I'm with comparable people. My game doesn't jump above my shoulders, but it's better.

Anonymous said...

So does the Flynn effect mean that older people really are blockheads?