[MarkPawelek (OP)]

Is it true that "98% of all variation in our educational attainment is non-genetic?"

I refer you to this commentary making the claim:
http://www.independentsciencenews.org/science-media/science-and-social-control-political-paralysis-and-the-genetics-agenda/

based on this recently published article:
GWAS of 126,559 Individuals Identifies Genetic Variants Associated with Educational Attainment
ScienceExpress / http://www.sciencemag.org/content/early/recent / 30 May 2013 / Page 11 / 10.1126/science.1235488
http://scholar.harvard.edu/files/laibson/files/gwas_science_053013.pdf

If it's the case why have the media ignored this news?

[alancalverd]

The claim seems bizarre. We know that some exceptional genetics, at the level we would describe as genetic disorder or anomaly, is associated with poor educational outcome, so the commonsense observation is that almost 100% of educability is genetically determined, but it happens that 95% of the population is close to the norm of whatever genetic structure determines it.

[evan_au]

Political biases can distort reports like this in either direction. It is easy to show (with biased arguments) that educational outcomes are almost 100% genetic or 100% environmental.

The reality as revealed by twin studies (identical vs fraternal twins; twins raised together or separately) is that many characteristics are about half environment vs genetic; some characteristics have a stronger genetic component than this. But even identical twins will have differences in de-novo mutations and differences in gene expression, which will hide some of the genetic factors.

Twin studies will indicate the degree of genetic vs environmental influence, but they won't tell you which genes are significant, or even whether any of the significant genes have a high probability of influencing the outcome being studied. For many complex characteristics (showing a "Normal" distribution), there are many influencing characteristics, each with a small impact. It takes a large sample size to detect even the largest of these individual small influences.

From skimming this paper, I deduce that they were not reading the actual genes that code for proteins, or even the control regions that control expression of the genes, but were looking for common "SNPs", which are short genetic changes in a segment of the population. It will take another generation of genetic studies to look at the underlying DNA.

While it takes a lot of genes to be "right" to get at the high end of educational accomplishment, it sometimes needs just one gene to be mutated to end up at the lower end of the bell curve. Identifying individual genetic changes which are deleterious will highlight areas of the genome which have a direct impact on educational achievement, and could form the basis for future studies looking at more subtle effects.

Another factor that I am sure must distort studies like this is the effects of genetic imprinting, where some genes inherited from the mother are disabled, and some other genes inherited from the father are disabled in the same individual. Some of the imprinted genes relate to brain development, which is very pertinent to studies of educational accomplishment. Earlier genetic screening techniques could tell what genes you carried, but not whether they were imprinted or not. This will halve the significance of any findings relating to imprinted genes.

Personally, I think that genetic studies will reveal more useful information as the costs come down, and larger studies using full-genome sequencing become possible (not just SNPs). 

• A lot of the advances of the 20th century came about because of the exponential growth implied by Moore's Law - a doubling of semiconductor performance about every 18 months.
  
• Genetic sequencing also is on an exponential trajectory - only the doubling time is somewhere around 3 to 6 months (depending on who you believe); if this rate can be sustained, genetic sequencing can be expected to make dramatic changes in the 21st century. I just hope that we learn to read DNA before we write.

As for tarring genetics with the tobacco brush, tobacco (and generally, breathing in smoke & pollution) is a bad move for your health & education.

[MarkPawelek (OP)]

Here are the supplementary materials.
http://www.sciencemag.org/content/suppl/2013/05/29/science.1235488.DC1/Rietveld.SM.revision.1.pdf. 160 pages excluding citations!

To answer the critics above.
1. They seem to discount previous research

To date, however, few if any robust associations between specific genetic variants and social-scientific outcomes have been identified likely because existing work has relied on samples that are too small.

2. They acknowledge that this study is out of skelter with received wisdom

Estimates suggest that around 40% of the variance in educational attainment is explained by genetic factors.

[MarkPawelek (OP)]

I bumping this because my detractors posted no counter evidence.

Science article: http://www.research.ed.ac.uk/portal/files/9199265/GWAS_of_126_559_Individuals_Identifies_Genetic_Variants_Associated_with.pdf

Supplementary Materials (171 pages): http://www.sciencemag.org/content/suppl/2013/05/29/science.1235488.DC1/Rietveld.SM.revision.1.pdf

They claim that 2% of educational attainment is due to genes.

alancalverd says the claim seems bizarre but posted no alternative evidence.

evan_au defers to 'twin studies', but posts no links: 

Political biases can distort reports like this in either direction. It is easy to show (with biased arguments) that educational outcomes are almost 100% genetic or 100% environmental.

I don't believe evan_au. If it's easy to show why did so many researchers agree on 2%. Why stick their necks out by quoting 2% when the prevailing orthodoxy says 50%?

Over 100 researchers took part in this study of 125000 individuals.  It looks like a massive study to me.

PS: The only Twin study I could find is from 23 years ago.
http://web.archive.org/web/20120227061723/http://www.psych.umn.edu/courses/spring05/hicksb/psy3135/bouchard_1990.pdf

It claimed that 70% of educational attainment was due to genetic factors. Elsewhere people have flatly told me the number is 50% and they seem quite certain about it.

The evidence says 2%, but scientists are certain that it's actually 50%.  Is it another case of science progressing one funeral at at time?  I hope not.

Just out of interest. How long before we're able to identify the specific genes which code for most of the difference in educational attainment. Which genes? Which mutations? What might the incidence of mutation be? How long before we can agree on whether it's 50% or 2%?

[alancalverd]

The first problem is to define "educational attainment" in a repeatably measurable form.

The second problem is to define what you mean by a 2% variation in attainment, and show that your measurements are sufficiently accurate to detect this.

The third problem is to test it over the entire range of human genetic variants, correcting for environmental and cultural influences.

The fourth problem is to relate your measurements to the everyday observation that, for instance, Downs Syndrome variants in any society tend not to gain many educational qualifications, however hard they try. 

Frankly, I think the quoted measure "failure to complete high school" is so dependent on social factors as to mask any possible genetic component. But there's the weasel word: the study was about attainment, not capability.

So here's a comparable experiment. Only about 1 person in 100 has a Grade 8 certificate in piano playing. Huge differentiation in attainment between amateur experts and the general population. But at first glance anyone with two hands could learn to play the piano, so we might usefully seek a genetic differentiation in capability to explain the small number of high achievers. (Unless you are genetically unable to play the piano at all due to, say, not having a right hand, in which case there is an obvious 100% genetic component.) But whatever we find will be completely overshadowed by the social factor of not having a piano. 

[MarkPawelek (OP)]

Once thing I noticed about this study was that they had narrowed it to 3 specific SNPs. I'm shocked they only found 3. I guess the effect of these 3 far outweighs effects from the rest of the genome. I'd assume that the next stage is to go after the genes, to discover the specific physiological pathways and to figure out how and why some of us are better at learning and problem solving. This is at the heart of the IQ debate that's been so fiercely debated all my life.

You've raised 4 huge problems there.  Yes. There definition of educational attainment is clearly too crude.

As for piano playing. Not everyone has the opportunity to learn. For those that do, piano students spend vastly different amounts of time practising. They fall into 2 main groups. Those spending minimum time and those voluntarily spending more time. The latter group have a much better chance of becoming professional musicians.

[alancalverd]

I never understood the "fierce debate" business. Debating the significance of an IQ test is like debating the number of angels on a pinhead.

In the real world, if you want to select candidates for training or employment, you set a specific aptitude test for the intended task, or pre-filter by requiring prior certification in task-related competences. Having an IQ of 160 doesn't mean you can kick a football, and having a degree in theoretical physics doesn't mean you can't, but a few winners' medals might indicate suitability for the job.

It would be a pretty unusual concert promoter who hired a guy who had never owned a piano, let alone been coached to a high level,  to lead a concerto, but remarkably few such would-be musicians complain about social discrimination. Satisfaction comes from doing what you can with what you've got. 

[MarkPawelek (OP)]

If you look at it in the context of UK eduction policy post-1945 to late 1960s it makes sense. At secondary level there were supposedly Grammar schools, Technical schools and Secondary Moderns. Children were selected for these at age 11. Kids at Grammar School got by far the best education, but no more than 25% of children were allowed in. My Secondary Moderns were academically crap. This was all justified on the grounds that only the best could benefit from a good education because they had better natural ability. That, and constant attempts to justify IQ as some kind of natural variation with a 'Bell curve', normal distribution. It was like an echo of 'The Mismeasure of Man' http://en.wikipedia.org/wiki/The_Mismeasure_of_Man 

[alancalverd]

All depends on what you mean by "best" and a "good education". AFAIK Germany has four or five different types of secondary school and a craft or technical education is valued: result? a very strong manufacturing industry paying good wages to skilled technicians, compared with the UK where the biggest economic sector is selling houses to one another, 51% of the population has a "university degree" (most of which would not qualify the holder for entrance into a proper university), 25% graduate unemployment, and no industry. 

The simple fact is that we need very few genuine graduates and desk-jockeys in real life, and the few that are suited to the requisite educational process actually do better if educated selectively. But the more important fact is that everyone else needs serious, high quality vocational training, which is expensive if done properly.   

[cheryl j]

If you look at it in the context of UK eduction policy post-1945 to late 1960s it makes sense. At secondary level there were supposedly Grammar schools, Technical schools and Secondary Moderns. Children were selected for these at age 11. Kids at Grammar School got by far the best education, but no more than 25% of children were allowed in. My Secondary Moderns were academically crap. This was all justified on the grounds that only the best could benefit from a good education because they had better natural ability. That, and constant attempts to justify IQ as some kind of natural variation with a 'Bell curve', normal distribution. It was like an echo of 'The Mismeasure of Man' http://en.wikipedia.org/wiki/The_Mismeasure_of_Man 

Frankly I think it is a huge mistake to direct children into certain areas or occupations based on measures of IQ or intelligence rather than interest. Friends I know who ended up in the building trades and construction because they supposedly were poor students use more math, more problem solving skills, more creative thinking than many people I know with University liberal arts degrees who ended up in offices performing the same clerical task again and again and again for wages slightly above what they would make at a fast food restaurant. It's a different world now.

[CliffordK]

Here is an interesting summary of two meta studies of identical twins reared apart.  If you are truly interested in the question, it might be worthwhile to acquire the books.

Unfortunately the answers of Nature vs Nurture seem to be about as clear as mud.  There apparently are a number of biases in the twin studies including age of separation, selection biases, and whether the twins are truly independent.  Also, the social status of the adoption families may not vary as much as normal community variance. 

Some of the studies apparently also concluded that identical twins would otherwise be identical in an identical environment, but there are often striking differences between identical twins (and even conjoined twins).

Personally I would say that I have a level of curiosity and intensity that has guided myself through my studies that not everyone possesses. 

For my SAT exam, my Math and Verbal scores were significantly different.  I'm trying to think if there is anything I could have done to flip them, and I'm dubious that it would have been possible.

[MarkPawelek (OP)]

I think when we've begun to identify which genes do what we'll be closer to knowing. That seems decades away.

GWAS studies seem to estimate too low (2%). GCTA studies have potential flaws too: Still Chasing Ghosts: A New Genetic Methodology Will Not Find the "Missing Heritability"

[MarkPawelek (OP)]

Childhood intelligence is heritable, highly polygenic and associated with FNBP1L
Mol Psychiatry. 2014 Feb;19(2):253-8. doi: 10.1038/mp.2012.184. Epub 2013 Jan 29.

aggregate effects of common SNPs explain 22-46% of phenotypic variation in childhood intelligence

But also ... IQ tests used may have major problems with their design:
The Heritability of Intelligence: Not What You Think

On the Nature and Nurture of Intelligence and Specific Cognitive Abilities: The More Heritable, the More Culture Dependent
DOI: 10.1177/0956797613493292
Psychological Science published online 8 October 2013
Kees-Jan Kan, Jelte M. Wicherts, Conor V. Dolan and Han L. J. van der Maas