Monday 16 March 2015

Genes for class, education and IQ

Barely a week passes until another paper gets published in which researchers delve into genetic material and find associations with higher level human behaviours which have always been of great interest: the class or status a person achieves in society; the extent to which they have learned some of the collective wisdom of society; and their ability to solve problems on their own.

Frankly, I find these associations between DNA and human behaviours pretty surprising. It was not the way I was brought up. My early education, as far as I can recall it, hammered in the notion that most things in life depended on effort. Teachers argued that the diligent application of one’s self to set tasks would result in greater skills, and the building of character. From this perspective the real school motto was: Do your homework. The most savage judgment meted out by this Spartan system to miscreants was that they were: “steeped in self pity and lazy to the core”.  Your self was not considered a subject of interest, only your results.

My tertiary education took on a more sociological flavour. Of course some effort was required, the Professors argued, but if the class structures were not conducive, then all your efforts could come to naught. Class had a big effect on your life chances, and it was all very unfair. Hence the impact, as I have already described, of my being offered a university foundation year which itself had grown out of the Workers’ Educational Association, offering adult education to the British working class, based on interest and merit, not the capricious circumstances of class allocation. Finding many bright people in working class occupations who had not had the benefit of an education, many of these professors were tempted to argue that the only difference between social classes was due to the unmerited accumulation of wealth and social connections by exploitative practices. Given that assumption, then many things follow, among which is the assumption that it is always legitimate to “control” for social class as a handicapping variable, in the sense that there cannot really be any difference between classes other than the hand of fate. Therefore, poking about in the wriggly, squishy reproductive fluids would not be likely to have anything to say about the class structure of society.

E Krapohl and R Plomin (2015) Genetic link between family socioeconomic status and children’s educational achievement estimated from genome-wide SNPs.  Molecular Psychiatry (2015), 1–7

 http://www.nature.com/mp/journal/vaop/ncurrent/pdf/mp20152a.pdf

Krapohl and Plomin say in their abstract: One of the best predictors of children’s educational achievement is their family’s socioeconomic status (SES), but the degree to which this association is genetically mediated remains unclear. For 3000 UK-representative unrelated children we found that genome-wide single-nucleotide polymorphisms could explain a third of the variance of scores on an age-16 UK national examination of educational achievement and half of the correlation between their scores and family SES. Moreover, genome-wide polygenic scores based on a previously published genome-wide association meta-analysis of total number of years in education accounted for ~ 3.0% variance in educational achievement and ~ 2.5% in family SES. This study provides the first molecular evidence for substantial genetic influence on differences in children’s educational achievement and its association with family SES.

I do not know if you share my surprise that educational achievement and social class can now be linked to the molecular level. We live in interesting times. Moving onwards, there is a nexus between genes, class and attainments which it would be good to understand. Selection for intelligence makes sense to me, but selection for class takes a little more time to sink in.

Here we report the first investigation of genetic influence on the variance of children’s educational achievement using DNA alone. The same DNA-based methods can also be used to estimate genetic influence on the covariance between traits. This enabled us to investigate possible genetic mediation of the best predictor of children’s educational achievement, their family’s SES. This correlation is often interpreted causally as family SES causing differences in children’s educational achievement.20 However, it remains unclear whether and to what extent the association between family SES and children’s educational achievement is genetically mediated, because twin and family research is limited to studying phenotypes that can vary within a family. Key aspects of children’s environment such as poverty, parental education and neighbourhood cannot be investigated using the twin method because it is methodologically impossible to decompose variance in phenotypes shared within twin pairs.

The DNA-based technique, genome-wide complex trait analysis fits the effects of genome-wide single-nucleotide polymorphisms as random effects in a mixed linear model to estimate variance or covariance captured by all SNPs simultaneously. Contrary to traditional family-based methods that estimate the genetic contribution to phenotypic variation or co-variation by known kinship coefficients, GCTA relies on empirical genetic resemblance established from identity by state inferred from genome-wide SNP similarity of ‘unrelated’ individuals.

Our GCTA results show that SNPs that are associated with both family SES and GCSE scores account for about half of the phenotypic correlation between SES and GCSE. Mediation analysis suggests that about one-third of this genetic effect also extends to children’s intelligence, but two-thirds of the genetic association between family SES and GCSE scores is independent of intelligence. In GPS analysis, we show that SNPs associated with total years of education in adulthood discovered by an independent large GWA meta-analysis13 explain up to 3% of the variance in children’s educational achievement in our sample, and up to 2% of the variance after controlling for intelligence.

The GCTA heritability estimate of 31% for children’s performance on a UK national examination at the end of compulsory education corroborates the vast literature of traditional family based methods, mostly the twin method, showing that variation in children’s educational achievement is under substantial genetic influence,4,5,7–9,45,46 with heritability estimates converging at ~ 50%. This commonly observed discrepancy in phenotypic variance explained by pedigree-based methods (that is, twin and family) and population-based methods (that is, GCTA) occurs because GCTA only captures genetic variance contributed by additive effects of common SNPs that are in sufficient linkage disequilibrium with the causal DNA variants.

An interesting note of caution emerges about the power of intelligence:

we find that children’s intelligence accounts for about one-third of the GCTA association between family SES and children’s educational achievement. However, it is interesting that two-thirds of the GCTA association is not accounted for by children’s intelligence. This finding of intelligenceindependent shared genetic variance between family SES and children’s educational achievement suggests that differences in educational achievement at the end of compulsory education and the level of education and occupation attained in adulthood are not merely the manifestation of differences in intelligence. This is in line with twin research that suggests that the heritability of educational achievement reflects many genetically influenced traits such as personality and self-efficacy, not just intelligence.

 

image

In summary, different ways of looking at the genetic code are beginning to provide ways of generating small but interesting predictions about complex behaviour, which very probably will get stronger either as more methods of analysis are employed, or when someone comes up with a theory which simplifies the number of comparisons to be made. Will you, dear reader, come up with the next step?

21 comments:

  1. What's "self-efficacy"? Not having your thumb up your bum?

    ReplyDelete
  2. Morale. Persistence. Character.

    ReplyDelete
  3. "My early education, as far as I can recall it, hammered in the notion that most things in life depended on effort." One of my teachers was more realistic.

    "Still finding everything comes easy?"

    "Aye, sir"

    "Lucky bugger."

    ReplyDelete
  4. You had kinder teachers than I remember!

    ReplyDelete
  5. This finding makes sense. IQ is not every thing. SES, Educational achievement depend on multiple factors including intelligence, self-control, future orientation, which might be all genetically based. We need to consider SES like a table with intelligence as one of supporting legs. Without it, the table will collapes. But the table also need other supporting legs (self-controls ect) to stand.

    IC

    ReplyDelete
    Replies
    1. Good metaphor.

      Delete
    2. Could be, but in studies of scholastic attainment with full population samples IQ is the strongest by far. We keep hoping that personality variables will account for additional variance, but the results are not overwhelming.

      Delete
  6. How do you measure characteristics such as honesty, compliance, diligence, charm, probity, sense of humor, care for others, ability to influence others, sense of time, etc? It seems one can set up tests for intelligence, but it's hard to put a numerical value on other personal qualities, especially the qualities we think are undesirable.

    Assuming such analyses as the paper mentioned in your post can be replicated, I would be interested to know if the genetic makeup of the highest SES group(s) changed with time. The descendants of feudal knights---do they hold similarly exalted positions in today's society? Did the Industrial Revolution change the relative SES of families holding distinctive genetic packages? Is the advent of computing machinery ushering in other changes?

    I am very wary of such research being badly applied by people who don't understand it. It would be too tempting for some to take it as endorsing the status quo.

    ReplyDelete
    Replies
    1. Honesty: testable, and currently tested but without a gold standard; compliance testable and frequently tested; charm testable, probity see honesty; sense of humour tested; care/altruism tested, influence testable; sense of time tested frequently. All of these can be shown with numerical values. SES hold up pretty well across centuries, Gregory Clark argues.

      Delete
    2. Gregory Clark uses family surnames, though. Given a certain degree of cuckoldry over the ages, surnames and genetic inheritance might not stem from the same family. (Witness the recent news stories about Richard III's remains.)

      As to SES, primogeniture would give the oldest male heir the greatest share of family wealth. It would be important to find out what happened to the descendants of the other siblings. The line which starts out with the greatest share of material wealth, and maintains that advantage with each generation, would have the best chance to maintain SES over generations. I would assume the oldest son would receive the lion's share of education, as well. If I start out a game of Monopoly with Park Place and Boardwalk, and an overly generous share of the money, I have better than even chances of winning the game.

      Delete
    3. @East Coast

      "Gregory Clark uses family surnames, though. Given a certain degree of cuckoldry over the ages, surnames and genetic inheritance might not stem from the same family."

      Genetic analyses show that historic non-paternity rates are on the order of 1-2% in most European societies.

      Delete
    4. Which would compound over generations. Unless, of course, there are clusters. If some female family lines are more likely to bear children within marriage, but not the husband's children, the compounding effect would be less than if any female had a 1-2% chance of bearing such a child.

      Delete
  7. "How do you measure characteristics such as ...": that's a fair question. But no doubt people once asked "How do you measure characteristics such as intelligence?" IQ testing is the best answer to date. I suppose the floor is open for anyone who wants to devise tests for the items on your list.

    ReplyDelete
  8. What explains the other 2/3? Is it physical or psychological traits which can be measured through objective measurements, or is it something like having the right mix of pheromones? What if it's something like, smells trustworthy to those who control capital and resources? Smells like a member of my tribe? As pheromones are a physical product of the body, a person's individual mixture may reflect genetic factors. Must reflect it, even.

    I wonder if in-person interviews are more likely to hire people of similar ancestry (even if unknown to the participants) than Skype interviews?

    ReplyDelete
    Replies
    1. All those things are possible. I think that attractiveness is probably the strongest additional factor: pretty face and body with good (not exaggerated) proportions is probably an additional boost, at least in getting hired, and probably in getting promoted. Unclear how much of the variance it accounts for.

      Delete
  9. "Weight balance" measure well our weight but don't measure the mass index, body mass distribution and/ or Metabolism, etc. Iq measure ( slightly) well our intelligence but don't measure smart "personality traits", creativity, holistic perception of reality (social, cultural, political, sexual and congenital... Maybe) etc.. Dynamics need a real scenario. Dynamic intelligence in real time can be measure slightly well by their readers. Something is wrong with my neighbohood or my country, something with my "survivability fitness".

    But is completely possible we have "smart" sillies, with good vocabulary and severe perceptual ability. Assynchronic development is not only a gifted trait because is universal and leftoid have this syndrome as well hardcore rightoids, impared behavioral plasticity and acceptance this plasticity or uncommon behaviors.


    Santoculto

    ReplyDelete
    Replies
    1. * severe DEFICITS in perceptual ability for leftomaniacs.

      No there "genes for social classes". There are phenotypical combinations which are contextually advantageous in today complex human societies.

      Santoculto

      Delete
  10. I think, my early education was effected from my hard working. I can say that it was hammered by the teachers badly.
    Psychologist in Australia | Children Psychologist

    ReplyDelete
  11. the stupid poor kid25 March 2015 at 01:07

    aight so look, a genetic correlation (gonna call it a 'genny co' from now on, to save time innit) is trying to tell you something interesting - to what extent are the genetic effects that influence trait A also influencing trait B in the same direction with the same magnitude of effect.

    in general this is quite an opaque metric. sadly, in this case, it aint telling you nothing solid about causality or shared biological mechanism underlying the two traits.

    lemme give you an example. let's say that everyone with high intelligence gets to marry someone who is richer, and low intelligence people marry people who are poorer. let's also say there is a genetic contribution to intelligence and that poor people and rich people don't mix too much. nothing too unrealistic about these assumptions right? ok so what we have is some very subtle population stratification in our study now - slightly different allele frequencies across the entire genome due to a bit of historical sub-structure generated by non-random mating. this creates an inflated genetic component to the variance in socioeconomic status. sweet.

    now let's take all the kids from all that romance and use bivariate gcta analysis to estimate a genny co between our two important and well defined traits (heh heh). wow! a significant correlation! the clever rich kids who inherited high intelligence alleles from one parent inherited rich-people-ancestry informative markers from the other parent, and vice versa with the stupid poor kids.

    let's seed this whole observational association by performing the study on a sample from a society which a) defines intelligence in some way (like do you get good grades in school), b) rewards individuals for the rest of their lives that exhibit characteristics that fit the definition that it landed on in (a), and c) has precious little social mobility. what exactly should we infer when our REML engine converges on a fit that some proportion of this autocorrelation is due to shared genetic variation?

    you know, on the estate we got a bunch of words to describe an association like this, but here is the one that is most pertinent: "confounded".

    ReplyDelete
  12. I think it interesting that a UK 2014 twin study found 62% of the differences among their scores on a significant test given at age 16 were due to genetic factors. They didn't take genetic samples, however.
    Even though the referenced study that found 31% on the same test took genetic samples, the researchers did not necessarily show that genetics were the cause of the effects.
    A better way to estimate the relative influences of genetics and environmental factors is to measure the degree of DNA methylation of genes. Although this technique is new, it isn't unreasonable to expect that studies performed in 2014 and 2015 would apply this technique if they intend to make findings that differentiate the contributions of genetic and epigenetic factors to human development and behavior.
    Here's the 2014 study as I curated it:
    http://surfaceyourrealself.com/2015/05/16/dividing-genetic-and-environmental-factors-when-assessing-educational-achievement-surfaceyourrealself/

    ReplyDelete