The Unz Review - Mobile
A Collection of Interesting, Important, and Controversial Perspectives Largely Excluded from the American Mainstream Media
 TeasersiSteve Blog
"Genetic Analysis of Social-Class Mobility in Five Longitudinal Studies"
🔊 Listen RSS
Email This Page to Someone

 Remember My Information



=>

Bookmark Toggle AllToCAdd to LibraryRemove from Library • BShow CommentNext New CommentNext New Reply
Search Text Case Sensitive  Exact Words  Include Comments
List of Bookmarks

From PNAS:

Genetic analysis of social-class mobility in five longitudinal studies

Daniel W. Belsky, Benjamin W. Domingue, Robbee Wedow, Louise Arseneault, Jason D. Boardman, Avshalom Caspi, Dalton Conley, Jason M. Fletcher, Jeremy Freese, Pamela Herd, Terrie E. Moffitt, Richie Poulton, Kamil Sicinski, Jasmin Wertz, and Kathleen Mullan Harris
PNAS July 9, 2018. 201801238; published ahead of print July 9, 2018.https://doi.org/10.1073/pnas.1801238115

Significance
Genome-wide association study (GWAS) discoveries about educational attainment have raised questions about the meaning of the genetics of success. These discoveries could offer clues about biological mechanisms or, because children inherit genetics and social class from parents, education-linked genetics could be spurious correlates of socially transmitted advantages. To distinguish between these hypotheses, we studied social mobility in five cohorts from three countries. We found that people with more education-linked genetics were more successful compared with parents and siblings. We also found mothers’ education-linked genetics predicted their children’s attainment over and above the children’s own genetics, indicating an environmentally mediated genetic effect. Findings reject pure social-transmission explanations of education GWAS discoveries. Instead, genetics influences attainment directly through social mobility and indirectly through family environments.

Abstract
A summary genetic measure, called a “polygenic score,” derived from a genome-wide association study (GWAS) of education can modestly predict a person’s educational and economic success. This prediction could signal a biological mechanism: Education-linked genetics could encode characteristics that help people get ahead in life. Alternatively, prediction could reflect social history: People from well-off families might stay well-off for social reasons, and these families might also look alike genetically. A key test to distinguish biological mechanism from social history is if people with higher education polygenic scores tend to climb the social ladder beyond their parents’ position. Upward mobility would indicate education-linked genetics encodes characteristics that foster success. We tested if education-linked polygenic scores predicted social mobility in >20,000 individuals in five longitudinal studies in the United States, Britain, and New Zealand. Participants with higher polygenic scores achieved more education and career success and accumulated more wealth. However, they also tended to come from better-off families. In the key test, participants with higher polygenic scores tended to be upwardly mobile compared with their parents. Moreover, in sibling-difference analysis, the sibling with the higher polygenic score was more upwardly mobile. Thus, education GWAS discoveries are not mere correlates of privilege; they influence social mobility within a life. Additional analyses revealed that a mother’s polygenic score predicted her child’s attainment over and above the child’s own polygenic score, suggesting parents’ genetics can also affect their children’s attainment through environmental pathways. Education GWAS discoveries affect socioeconomic attainment through influence on individuals’ family-of-origin environments and their social mobility.

 
Hide 34 CommentsLeave a Comment
Commenters to Ignore...to FollowEndorsed Only
Trim Comments?
    []
  1. Alfa158 says:

    “Education linked genetics” and “polygenic scores”. That would be genetics that affect IQ but the researchers don’t want to be hounded out of their jobs and kicked out of family get togethers? I’m sure they aren’t going full Lysenko on us and implying that education can affect your genetics, so they must mean those genetic traits that make you able to achieve even more educationally than other members of your family, social and economic groups with the same environmental advantages.

    Read More
    • Replies: @utu
    Actually you are wrong. So far predictions of IQ with some polygenic score could account for no more than 7% of variance while the prediction of education attainment in this paper and in Steven Hsu paper explains 9% of variance. So, if they concentrated on IQ instead of the education attainment their results would be less impressive. After they removed parents' education it is less than 9%.
    ReplyAgree/Disagree/Etc.
    AgreeDisagreeLOLTroll
    These buttons register your public Agreement, Disagreement, Troll, or LOL with the selected comment. They are ONLY available to recent, frequent commenters who have saved their Name+Email using the 'Remember My Information' checkbox, and may also ONLY be used once per hour.
    Ignore Commenter Follow Commenter
    More... This Commenter This Thread Hide Thread Display All Comments
  2. Polynikes says:

    Anything to avoid saying those magical but odious two little letters, huh?

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  3. Hubbub says:

    …can modestly predict…

    Says about all that needs to be said.

    Read More
    • Replies: @Anon
    Polygenic scores use currently known genes, so as time goes on and more genes comes to light prediction should improve.
    , @utu
    Very, very modestly.

    Prediction of General Certificate of Education Examination (GCSE) level (none, 1, 2, or 3) with polygenic score (PGS)

    r(GCSE,PGS) = 0.27 (7.3% of variance)

    r(GCSE,PGS, parental-education adjusted) =0.16 (2.6% of variance)


    I could not find how many SNPs actually contributed to the polygenic score.

    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  4. PNAS? Shouldn’t they all be male?

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  5. The study fails to define success in any other way than attaining college degrees? Only overeducated lunk heads trying to rationalize needing to be incubated for nearly 30 years at great expense would generate such a confounded! hypothesis. Silly wastrels coming up with bad ideas. Nothing to see here.

    Read More
    • Replies: @AndrewR
    What? They're saying "polygenic scores" correlate with educational attainment and wealth. But we already knew that educational attainment and wealth correlate. Yes, one could reasonably object to defining "getting ahead in life" as having more formal education and more wealth, but that's how most people would probably define it anyway.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  6. Off topic, but the latest hate hoax involves a black millionaire finding a note in his hotel room with the words “You’re a N—-r” written on it:

    https://nypost.com/2018/07/07/african-american-millionaire-finds-note-with-racial-slur-in-hotel-room/

    Read More
    • Replies: @Anon
    We all know he wrote the word himself.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  7. utu says:
    @Alfa158
    “Education linked genetics” and “polygenic scores”. That would be genetics that affect IQ but the researchers don’t want to be hounded out of their jobs and kicked out of family get togethers? I’m sure they aren’t going full Lysenko on us and implying that education can affect your genetics, so they must mean those genetic traits that make you able to achieve even more educationally than other members of your family, social and economic groups with the same environmental advantages.

    Actually you are wrong. So far predictions of IQ with some polygenic score could account for no more than 7% of variance while the prediction of education attainment in this paper and in Steven Hsu paper explains 9% of variance. So, if they concentrated on IQ instead of the education attainment their results would be less impressive. After they removed parents’ education it is less than 9%.

    Read More
    • Replies: @res
    Do you have the sample sizes for those two studies handy? I am curious how the power compares. I think EA is a much easier to obtain phenotype than IQ. It is also far less controversial.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  8. res says:
    @utu
    Actually you are wrong. So far predictions of IQ with some polygenic score could account for no more than 7% of variance while the prediction of education attainment in this paper and in Steven Hsu paper explains 9% of variance. So, if they concentrated on IQ instead of the education attainment their results would be less impressive. After they removed parents' education it is less than 9%.

    Do you have the sample sizes for those two studies handy? I am curious how the power compares. I think EA is a much easier to obtain phenotype than IQ. It is also far less controversial.

    Read More
    • Replies: @utu
    the 7% IQ study was discussed here and by James Thompson.

    the 9% EA by Hsu is in his paper that also was discussed by James Thompson.

    I think samples where pretty large but I do not remember exact numbers.
    , @Alfa158
    I’m trying to learn a little about genetics. If I have this right a polygene is a set of genes that together control a trait. If I’m following utu’s explanation, there is a polygene which correlates to higher educational achievement, and it predicts individual achievement more accurately than the polygene that correlates to IQ test scores.
    I read the paper linked and the researchers seem to be concluding that educational achievement, financial success and social status are linked to both the person’s family and social environment but also to an identifiable polygene for educational aptitude. Children with a higher “score” on this polygene will outperform parents or siblings who come from the same environment but score lower on this polygene.
    I haven’t had a chance yet to dig in on exactly what they are measuring or how they are measuring it in order to arrive at a polygene score. A lot of the reference links in the paper come up as inaccessible to my browser.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  9. anon[118] • Disclaimer says:

    I need to run this thing through Google Translate…to English.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  10. utu says:
    @res
    Do you have the sample sizes for those two studies handy? I am curious how the power compares. I think EA is a much easier to obtain phenotype than IQ. It is also far less controversial.

    the 7% IQ study was discussed here and by James Thompson.

    the 9% EA by Hsu is in his paper that also was discussed by James Thompson.

    I think samples where pretty large but I do not remember exact numbers.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  11. Off topic, but up the Steve Sailer alley:

    Tiny male monkeys use stone tools to break open food.

    Curiously, only male monkeys were seen wielding stone tools, even though females were often foraging nearby. This can’t be explained by females’ smaller size, given that juvenile males were able to use tools.

    I am proud of the WaPo for allowing mention of the male part of this and for refuting an easy explanation and without offering assurance that more study will be needed. Perhaps this was acceptable because the author had already established that these monkeys live in matrilineal troops and because the otherwise hatefact was not mentioned until the ninth overall paragraph.

    I imagine that when if the NYT ever gets around to writing about these monkeys, they will wait until a more respectable 15 paragraphs to drop that one in (if it gets mentioned) and that they will reassure us with some combination of A. no female has been seen using tools so far, B. the females are the leaders of these troops and so they don’t need extra food like these puny males (i.e. they could do the same thing but because they are superior they don’t have to), and C. build a nice strawman to show how unrelated we are to these creatures and that they did not play into our evolution, and therefore the idea that this proves anything is clearly false.

    https://www.washingtonpost.com/news/speaking-of-science/wp/2018/07/06/these-tiny-monkeys-have-entered-their-stone-age-with-a-bang/?

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  12. Anon[183] • Disclaimer says:
    @Hubbub

    ...can modestly predict...
     
    Says about all that needs to be said.

    Polygenic scores use currently known genes, so as time goes on and more genes comes to light prediction should improve.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  13. Alfa158 says:
    @res
    Do you have the sample sizes for those two studies handy? I am curious how the power compares. I think EA is a much easier to obtain phenotype than IQ. It is also far less controversial.

    I’m trying to learn a little about genetics. If I have this right a polygene is a set of genes that together control a trait. If I’m following utu’s explanation, there is a polygene which correlates to higher educational achievement, and it predicts individual achievement more accurately than the polygene that correlates to IQ test scores.
    I read the paper linked and the researchers seem to be concluding that educational achievement, financial success and social status are linked to both the person’s family and social environment but also to an identifiable polygene for educational aptitude. Children with a higher “score” on this polygene will outperform parents or siblings who come from the same environment but score lower on this polygene.
    I haven’t had a chance yet to dig in on exactly what they are measuring or how they are measuring it in order to arrive at a polygene score. A lot of the reference links in the paper come up as inaccessible to my browser.

    Read More
    • Replies: @res
    I'm not sure of your background and these ideas have a lot of layers of complexity. I'll bring up a few things and see if any of them help.

    First, I'm not really used to the term "polygene", but it does look apropos: https://en.wikipedia.org/wiki/Polygene
    A simple definition: "a gene whose individual effect on a phenotype is too small to be observed, but which can act together with others to produce observable variation."

    Regarding the paper, they are working from a polygenic score (PGS). This is a weighted sum based on a number of SNPs (single nucleotide polymorphisms). Both the particular SNPs and their weights are determined by a GWAS (genome wide association study). PGS are often expressed as z-scores where 0 is the group average and 1 is the standard deviation.

    Genes and SNPs are different, though people are often sloppy with their terminology here (I think the polygene definition I gave above is an example of this sloppiness). Briefly, genes are long sequences of the genetic code that translate into proteins while SNPs are individual changes in the genetic code.

    This is a good overview of GWAS but I'm not sure it is at the right level for you: http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1002822
    Wikipedia: https://en.wikipedia.org/wiki/Genome-wide_association_study

    I think what utu was saying is that the PGS we have for EA (Educational Attainment) are slightly better predictors (of that) than the PGS we have for IQ are predictors of IQ.

    I think there is some truth to your description in comment 1 (utu and I disagree a fair bit here sometimes). EA and IQ tend to correlate highly and are fairly good proxies for each other, but they are not the same thing. EA data tends to be more easily available and less controversial.

    Something like this course might be worthwhile for getting a baseline understanding: https://www.coursera.org/lecture/genetics-evolution/introduction-to-genetics-g-kNR29
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  14. res says:
    @Alfa158
    I’m trying to learn a little about genetics. If I have this right a polygene is a set of genes that together control a trait. If I’m following utu’s explanation, there is a polygene which correlates to higher educational achievement, and it predicts individual achievement more accurately than the polygene that correlates to IQ test scores.
    I read the paper linked and the researchers seem to be concluding that educational achievement, financial success and social status are linked to both the person’s family and social environment but also to an identifiable polygene for educational aptitude. Children with a higher “score” on this polygene will outperform parents or siblings who come from the same environment but score lower on this polygene.
    I haven’t had a chance yet to dig in on exactly what they are measuring or how they are measuring it in order to arrive at a polygene score. A lot of the reference links in the paper come up as inaccessible to my browser.

    I’m not sure of your background and these ideas have a lot of layers of complexity. I’ll bring up a few things and see if any of them help.

    First, I’m not really used to the term “polygene”, but it does look apropos: https://en.wikipedia.org/wiki/Polygene
    A simple definition: “a gene whose individual effect on a phenotype is too small to be observed, but which can act together with others to produce observable variation.”

    Regarding the paper, they are working from a polygenic score (PGS). This is a weighted sum based on a number of SNPs (single nucleotide polymorphisms). Both the particular SNPs and their weights are determined by a GWAS (genome wide association study). PGS are often expressed as z-scores where 0 is the group average and 1 is the standard deviation.

    Genes and SNPs are different, though people are often sloppy with their terminology here (I think the polygene definition I gave above is an example of this sloppiness). Briefly, genes are long sequences of the genetic code that translate into proteins while SNPs are individual changes in the genetic code.

    This is a good overview of GWAS but I’m not sure it is at the right level for you: http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1002822
    Wikipedia: https://en.wikipedia.org/wiki/Genome-wide_association_study

    I think what utu was saying is that the PGS we have for EA (Educational Attainment) are slightly better predictors (of that) than the PGS we have for IQ are predictors of IQ.

    I think there is some truth to your description in comment 1 (utu and I disagree a fair bit here sometimes). EA and IQ tend to correlate highly and are fairly good proxies for each other, but they are not the same thing. EA data tends to be more easily available and less controversial.

    Something like this course might be worthwhile for getting a baseline understanding: https://www.coursera.org/lecture/genetics-evolution/introduction-to-genetics-g-kNR29

    Read More
    • Replies: @Alfa158
    Thanks for the links. My B.S. was in Physics, I went into the electronics engineering, but my graduate degree was in business instead of science and I ending up spending most of my career in marketing, so the hardcore science and math are very rusty.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  15. Anon[257] • Disclaimer says:
    @Cognitive Miser
    Off topic, but the latest hate hoax involves a black millionaire finding a note in his hotel room with the words "You're a N----r" written on it:

    https://nypost.com/2018/07/07/african-american-millionaire-finds-note-with-racial-slur-in-hotel-room/

    We all know he wrote the word himself.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  16. “I read the paper linked and the researchers seem to be concluding that educational achievement, financial success and social status are linked to both the person’s family and social environment but also to an identifiable polygene for educational aptitude. Children with a higher “score” on this polygene will outperform parents or siblings who come from the same environment”

    Somewhat on topic: And yet, in society, we’ve been told that this current generation is the first in US history to not outperform economically the generation of their parents. In other words, either the kids aren’t “scoring” as high, or for whatever the reason, as a generation, they won’t tend to make as much money as their parents did, much less surpass them and thus their financial success, compared to their parents generation, will be lesser. Perhaps it evens out with the fact that compared to the Baby Boom Generation, and Gen. X, the Millennials comprise a smaller generation (measured by the total number of persons born within the generation) as compared to previous generations, and thus there are fewer kids that can score higher.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  17. Educational attainment has a huge data-efficiency advantage over intelligence test score as the variable to explain. Most health studies which gather DNA data also ask participants about their educational attainment (usually captured by years of schooling) so there exist massive databases of linked DNA-EA observations which can be tapped. Using intelligence test scores requires administering an intelligence test to each participant providing DNA so the available databases are much smaller.

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.

    Read More
    • Replies: @Bill

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.
     
    Would you mind providing a link explaining why this is done? At first blush, this strikes me as bizarre.
    , @res

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.
     
    This is a key point. I wish researchers would publish the coefficients and variance explained for the controlled for PCs.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  18. utu says:
    @Hubbub

    ...can modestly predict...
     
    Says about all that needs to be said.

    Very, very modestly.

    Prediction of General Certificate of Education Examination (GCSE) level (none, 1, 2, or 3) with polygenic score (PGS)

    r(GCSE,PGS) = 0.27 (7.3% of variance)

    r(GCSE,PGS, parental-education adjusted) =0.16 (2.6% of variance)

    I could not find how many SNPs actually contributed to the polygenic score.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  19. dearieme says:

    “genetics influences attainment directly through social mobility and indirectly through family environments”: well, stone the crows! Knock me dahn wiv a fevva. Did you ever? Swipe me! Whooda thunk it? And other expressions of astonishment.

    It’s remarkable that scholars take their careers in their hands when they proclaim the bleedin’ obvious.

    On t’other hand, isn’t there a US scholar whose evidence suggests that the effect of family environment has generally dwindled away by the time one is middle-aged? I suppose that wouldn’t show up if one measures attainment as getting a degree. (I say “generally” because I think she excepted the effects of extreme and unusual family environments.)

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  20. AndrewR says:
    @miss marple
    The study fails to define success in any other way than attaining college degrees? Only overeducated lunk heads trying to rationalize needing to be incubated for nearly 30 years at great expense would generate such a confounded! hypothesis. Silly wastrels coming up with bad ideas. Nothing to see here.

    What? They’re saying “polygenic scores” correlate with educational attainment and wealth. But we already knew that educational attainment and wealth correlate. Yes, one could reasonably object to defining “getting ahead in life” as having more formal education and more wealth, but that’s how most people would probably define it anyway.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  21. Bill says:
    @Peter Johnson
    Educational attainment has a huge data-efficiency advantage over intelligence test score as the variable to explain. Most health studies which gather DNA data also ask participants about their educational attainment (usually captured by years of schooling) so there exist massive databases of linked DNA-EA observations which can be tapped. Using intelligence test scores requires administering an intelligence test to each participant providing DNA so the available databases are much smaller.

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.

    Would you mind providing a link explaining why this is done? At first blush, this strikes me as bizarre.

    Read More
    • Replies: @res
    It is done to control for population structure. The UK Biobank is a good example. They actually provide the first 20 PCs in their data.

    Much detail in this pdf, along with many interesting PCA plots of their data.
    http://www.ukbiobank.ac.uk/wp-content/uploads/2014/04/UKBiobank_genotyping_QC_documentation-web.pdf

    There is some discussion of controlling for population stratification in this presentation: http://faculty.washington.edu/tathornt/SISG2015/lectures/assoc2015session06.pdf

    Hopefully Peter Johnson can provide more detail. I would like to learn more about this myself.

    P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!).
    , @Peter Johnson
    My answer for Bill:

    Suppose that society includes a hidden hereditary aristocracy who help each other to get more years of schooling through environmental factors (such as admitting each others' children to better schools). Their common genetic signals might be mistaken for a "genetic" factor in educational attainment whereas in fact it is environmentally caused but genetically linked (not causally). Whitening out the 15 top commonalities using principal components eliminates any such genetically-linked but environmentally-causal factors. It is overkill IMHO, but does away with that potential criticism of polygenic scores.

    Yes your intuition is correct that it probably is too much, but genetic researchers studying educational attainment and intelligence go to enormous lengths to eliminate all potential sources of criticisms for their findings.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  22. Tiny Duck says:

    I think you need to watch adam ruins everything and read Leonard Pitts

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  23. res says:
    @Peter Johnson
    Educational attainment has a huge data-efficiency advantage over intelligence test score as the variable to explain. Most health studies which gather DNA data also ask participants about their educational attainment (usually captured by years of schooling) so there exist massive databases of linked DNA-EA observations which can be tapped. Using intelligence test scores requires administering an intelligence test to each participant providing DNA so the available databases are much smaller.

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.

    This is a key point. I wish researchers would publish the coefficients and variance explained for the controlled for PCs.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  24. res says:
    @Bill

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.
     
    Would you mind providing a link explaining why this is done? At first blush, this strikes me as bizarre.

    It is done to control for population structure. The UK Biobank is a good example. They actually provide the first 20 PCs in their data.

    Much detail in this pdf, along with many interesting PCA plots of their data.

    http://www.ukbiobank.ac.uk/wp-content/uploads/2014/04/UKBiobank_genotyping_QC_documentation-web.pdf

    There is some discussion of controlling for population stratification in this presentation: http://faculty.washington.edu/tathornt/SISG2015/lectures/assoc2015session06.pdf

    Hopefully Peter Johnson can provide more detail. I would like to learn more about this myself.

    P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!).

    Read More
    • Replies: @dearieme
    "P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!)."

    There's none so blind as those who cannot see.
    , @Bill
    Thank you.

    So, the idea is to remove / refuse to look at those PCs which correspond closely to "population structure" i.e. race. Figure 2 in that first link is pretty persuasive that the first 4 PCs have something to do with race. But how do we know that we are not throwing away variation which actually has something to do with intelligence or whatever other phenotypic characteristics go into educational attainment? If you believe that educational-attainment-conducing genotypes really are correlated with race, isn't this a recipe for underestimating their effects (because measurement error, aka classification error) is enhanced by "controlling" for too many things?
    , @utu
    Useful links.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  25. @Bill

    Keep in mind that these explained variance proportions are the proportion explained after removing the first 15 principal components. This is done to avoid any environmentally-linked genetic differences hidden in these principal components, but it also has the effect of lowering attainable explanatory power.
     
    Would you mind providing a link explaining why this is done? At first blush, this strikes me as bizarre.

    My answer for Bill:

    Suppose that society includes a hidden hereditary aristocracy who help each other to get more years of schooling through environmental factors (such as admitting each others’ children to better schools). Their common genetic signals might be mistaken for a “genetic” factor in educational attainment whereas in fact it is environmentally caused but genetically linked (not causally). Whitening out the 15 top commonalities using principal components eliminates any such genetically-linked but environmentally-causal factors. It is overkill IMHO, but does away with that potential criticism of polygenic scores.

    Yes your intuition is correct that it probably is too much, but genetic researchers studying educational attainment and intelligence go to enormous lengths to eliminate all potential sources of criticisms for their findings.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  26. Dr. Doom says:

    ACME Looniversity has found that IQ had little effect on students edumacashunull advancermentations.

    Sociology Majors and Minorities in Minority Grievance Departments were able to pass with rainbow colors irregardless of IQ or the ability to form complete grammatically correct sentences.

    Scallers are now demanding the removal of White subjects like Math, Science and that “book learning”.

    Edumacashunull advancermentation. The Future looks a lot like the Dark Ages.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  27. dearieme says:
    @res
    It is done to control for population structure. The UK Biobank is a good example. They actually provide the first 20 PCs in their data.

    Much detail in this pdf, along with many interesting PCA plots of their data.
    http://www.ukbiobank.ac.uk/wp-content/uploads/2014/04/UKBiobank_genotyping_QC_documentation-web.pdf

    There is some discussion of controlling for population stratification in this presentation: http://faculty.washington.edu/tathornt/SISG2015/lectures/assoc2015session06.pdf

    Hopefully Peter Johnson can provide more detail. I would like to learn more about this myself.

    P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!).

    “P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!).”

    There’s none so blind as those who cannot see.

    Read More
    • Replies: @res
    Don't you mean "...will not see"? The choice part is important.

    In any case, I strongly agree with you.

    P.S. Some background for the saying: http://www.actualfreedom.com.au/richard/abditorium/nonesoblind.htm
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  28. Bill says:
    @res
    It is done to control for population structure. The UK Biobank is a good example. They actually provide the first 20 PCs in their data.

    Much detail in this pdf, along with many interesting PCA plots of their data.
    http://www.ukbiobank.ac.uk/wp-content/uploads/2014/04/UKBiobank_genotyping_QC_documentation-web.pdf

    There is some discussion of controlling for population stratification in this presentation: http://faculty.washington.edu/tathornt/SISG2015/lectures/assoc2015session06.pdf

    Hopefully Peter Johnson can provide more detail. I would like to learn more about this myself.

    P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!).

    Thank you.

    So, the idea is to remove / refuse to look at those PCs which correspond closely to “population structure” i.e. race. Figure 2 in that first link is pretty persuasive that the first 4 PCs have something to do with race. But how do we know that we are not throwing away variation which actually has something to do with intelligence or whatever other phenotypic characteristics go into educational attainment? If you believe that educational-attainment-conducing genotypes really are correlated with race, isn’t this a recipe for underestimating their effects (because measurement error, aka classification error) is enhanced by “controlling” for too many things?

    Read More
    • Replies: @res

    But how do we know that we are not throwing away variation which actually has something to do with intelligence or whatever other phenotypic characteristics go into educational attainment?
     
    I don't think there is any way of knowing for sure. This is why I advocate for researchers making clear how much variance is being explained by the PC regression variables. That would at least allow us to detect whether or not there is something important going on there that needs explanation.

    Peter is very much correct about "genetic researchers studying educational attainment and intelligence go to enormous lengths to eliminate all potential sources of criticisms for their findings." Unfortunately (IMHO) they are as concerned about political correctness issues as they are about real methodological issues.

    I should also be clear that most GWAS already try to eliminate continent level racial variation up front. A typical GWAS based on the UK Biobank will only look at European ancestry individuals in the first place. This lessens the impact of controlling for the PCs. They also typically exclude related individuals.

    If you believe that educational-attainment-conducing genotypes really are correlated with race, isn’t this a recipe for underestimating their effects (because measurement error, aka classification error) is enhanced by “controlling” for too many things?
     
    IMHO, yes.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  29. utu says:
    @res
    It is done to control for population structure. The UK Biobank is a good example. They actually provide the first 20 PCs in their data.

    Much detail in this pdf, along with many interesting PCA plots of their data.
    http://www.ukbiobank.ac.uk/wp-content/uploads/2014/04/UKBiobank_genotyping_QC_documentation-web.pdf

    There is some discussion of controlling for population stratification in this presentation: http://faculty.washington.edu/tathornt/SISG2015/lectures/assoc2015session06.pdf

    Hopefully Peter Johnson can provide more detail. I would like to learn more about this myself.

    P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!).

    Useful links.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  30. Alfa158 says:
    @res
    I'm not sure of your background and these ideas have a lot of layers of complexity. I'll bring up a few things and see if any of them help.

    First, I'm not really used to the term "polygene", but it does look apropos: https://en.wikipedia.org/wiki/Polygene
    A simple definition: "a gene whose individual effect on a phenotype is too small to be observed, but which can act together with others to produce observable variation."

    Regarding the paper, they are working from a polygenic score (PGS). This is a weighted sum based on a number of SNPs (single nucleotide polymorphisms). Both the particular SNPs and their weights are determined by a GWAS (genome wide association study). PGS are often expressed as z-scores where 0 is the group average and 1 is the standard deviation.

    Genes and SNPs are different, though people are often sloppy with their terminology here (I think the polygene definition I gave above is an example of this sloppiness). Briefly, genes are long sequences of the genetic code that translate into proteins while SNPs are individual changes in the genetic code.

    This is a good overview of GWAS but I'm not sure it is at the right level for you: http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1002822
    Wikipedia: https://en.wikipedia.org/wiki/Genome-wide_association_study

    I think what utu was saying is that the PGS we have for EA (Educational Attainment) are slightly better predictors (of that) than the PGS we have for IQ are predictors of IQ.

    I think there is some truth to your description in comment 1 (utu and I disagree a fair bit here sometimes). EA and IQ tend to correlate highly and are fairly good proxies for each other, but they are not the same thing. EA data tends to be more easily available and less controversial.

    Something like this course might be worthwhile for getting a baseline understanding: https://www.coursera.org/lecture/genetics-evolution/introduction-to-genetics-g-kNR29

    Thanks for the links. My B.S. was in Physics, I went into the electronics engineering, but my graduate degree was in business instead of science and I ending up spending most of my career in marketing, so the hardcore science and math are very rusty.

    Read More
    • Replies: @res
    So technical, but rusty. I'm guessing no formal classes in genetics? If so, a lot like my background circa a few years ago. I've taken a number of classes since then both to shake the rust off a bit and expand my knowledge in a variety of areas (e.g. genetics and the underlying biology).

    If you are interested in this sort of paper, I think some background in Behavioral Genetics is useful. I enjoyed this class and found it very informative: https://www.coursera.org/learn/behavioralgenetics
    The instructor, Matt McGue, is at the University of Minnesota (home of a famous twin study) and was a coauthor of the Rietveld et al. 2013 educational attainment GWAS: http://science.sciencemag.org/content/340/6139/1467
    Here is his publication history: https://scholar.google.com/citations?user=KeMR2u0AAAAJ&hl=en

    If you have the time and inclination I highly recommend his class.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  31. res says:
    @Bill
    Thank you.

    So, the idea is to remove / refuse to look at those PCs which correspond closely to "population structure" i.e. race. Figure 2 in that first link is pretty persuasive that the first 4 PCs have something to do with race. But how do we know that we are not throwing away variation which actually has something to do with intelligence or whatever other phenotypic characteristics go into educational attainment? If you believe that educational-attainment-conducing genotypes really are correlated with race, isn't this a recipe for underestimating their effects (because measurement error, aka classification error) is enhanced by "controlling" for too many things?

    But how do we know that we are not throwing away variation which actually has something to do with intelligence or whatever other phenotypic characteristics go into educational attainment?

    I don’t think there is any way of knowing for sure. This is why I advocate for researchers making clear how much variance is being explained by the PC regression variables. That would at least allow us to detect whether or not there is something important going on there that needs explanation.

    Peter is very much correct about “genetic researchers studying educational attainment and intelligence go to enormous lengths to eliminate all potential sources of criticisms for their findings.” Unfortunately (IMHO) they are as concerned about political correctness issues as they are about real methodological issues.

    I should also be clear that most GWAS already try to eliminate continent level racial variation up front. A typical GWAS based on the UK Biobank will only look at European ancestry individuals in the first place. This lessens the impact of controlling for the PCs. They also typically exclude related individuals.

    If you believe that educational-attainment-conducing genotypes really are correlated with race, isn’t this a recipe for underestimating their effects (because measurement error, aka classification error) is enhanced by “controlling” for too many things?

    IMHO, yes.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  32. res says:
    @dearieme
    "P.S. Remember this when you read about how racial differences are not seen in various results (which have been controlled for exactly those PCs which correspond with biological race!)."

    There's none so blind as those who cannot see.

    Don’t you mean “…will not see”? The choice part is important.

    In any case, I strongly agree with you.

    P.S. Some background for the saying: http://www.actualfreedom.com.au/richard/abditorium/nonesoblind.htm

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  33. res says:
    @Alfa158
    Thanks for the links. My B.S. was in Physics, I went into the electronics engineering, but my graduate degree was in business instead of science and I ending up spending most of my career in marketing, so the hardcore science and math are very rusty.

    So technical, but rusty. I’m guessing no formal classes in genetics? If so, a lot like my background circa a few years ago. I’ve taken a number of classes since then both to shake the rust off a bit and expand my knowledge in a variety of areas (e.g. genetics and the underlying biology).

    If you are interested in this sort of paper, I think some background in Behavioral Genetics is useful. I enjoyed this class and found it very informative: https://www.coursera.org/learn/behavioralgenetics
    The instructor, Matt McGue, is at the University of Minnesota (home of a famous twin study) and was a coauthor of the Rietveld et al. 2013 educational attainment GWAS: http://science.sciencemag.org/content/340/6139/1467
    Here is his publication history: https://scholar.google.com/citations?user=KeMR2u0AAAAJ&hl=en

    If you have the time and inclination I highly recommend his class.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  34. Alfa158 says:

    Correct, one semester each of biology, inorganic chemistry, and organic chemistry. My B.S. was back in the 60’s so there was a whole lot less genetic science going on! That was so far back that I used a slide rule, mainframe computer with punch cards, and worked after hours on a research project in nuclear magnetic resonance before the nuclear part was dropped so medical patients wouldn’t freak.

    Oh, forgot to add; by cracky!

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
Current Commenter says:

Leave a Reply - Comments are moderated by iSteve, at whim.


 Remember My InformationWhy?
 Email Replies to my Comment
Submitted comments become the property of The Unz Review and may be republished elsewhere at the sole discretion of the latter
Subscribe to This Comment Thread via RSS Subscribe to All Steve Sailer Comments via RSS
PastClassics
The evidence is clear — but often ignored
The “war hero” candidate buried information about POWs left behind in Vietnam.
Talk TV sensationalists and axe-grinding ideologues have fallen for a myth of immigrant lawlessness.
The major media overlooked Communist spies and Madoff’s fraud. What are they missing today?