The Unz Review - Mobile
A Collection of Interesting, Important, and Controversial Perspectives Largely Excluded from the American Mainstream Media
 Gene Expression BlogTeasers
Razib Khan at the Center of Eurasia
Email This Page to Someone

 Remember My Information



=>
Search Text Case Sensitive  Exact Words  Include Comments

Razib_Khan
The Eurogenes blog is running a fundraiser. I chipped in mostly to support his continued blogging. I don’t agree with everything he posts, but the site is a good and valuable resource. “Genome blogging” hasn’t gotten as far as I’d have thought it would have in 2010, mostly because the initial burst of enthusiasm wasn’t followed up by a consistent producer community (in addition to the commentariat). But Eurogenes soldiers on….

0780e5c94be98d65256f715b26529e90 In any case, as part of the donation I got an analysis of my genotype. I wasn’t super interested in this because I know a fair amount about my genotype, and the analysis isn’t too informative for someone like me who is >10% East Eurasian. But I thought I would post the PCA because it’s interesting. It’s hard to see in the image above (click it for a larger version), but you notice that I am almost equally positioned between the antipodes of Eurasia (Western Europeans vs. East Asians). Most South Asians occupy a position in between, but definitely skewed toward West Eurasians (the green). But you notice I’m nearly the most “eastern” of the main cluster of South Asians. Depending on how you calculate it I’m between 10% and 20% East Eurasian. This “pulls” me in that direction more than almost all other South Asians. Like a colossus I look east, and I look west, and stand athwart Eurasia bridging the gap.

In any case, if you have $12 to spare, think about donating to Eurogenes. Logistics are at the link.

 
• Category: Science • Tags: Genomics 
Commenters to Ignore...to FollowEndorsed Only
    []
  1. Marcus says:

    Well Bengal is where the Mongolid and Caucasoid traits start to intermix, right? Of course phenotype =/ genotype.

    https://en.wikipedia.org/wiki/Chakma_people

    Coincidentally I recently read something like this http://www.hindustantimes.com/india-news/manipuri-woman-alleges-racism-harassment-at-delhi-airport-s-immigration-desk/story-yoylKApQWGqZRGJwjRNFQM.html

    Read More
    • Replies: @Razib Khan
    it's a pulse admixture happened 1 to 2 thousand years ago. but yeah. also, the south asian ancestry spills over into burma.
    , @vijay
    The entire northeastern population gets abused in Delhi. Correction, all women get abused in Delhi. Correction again, the abuse starts when the train from Assam enters Bihar. Something about women wearing pants, etc. etc. Bimaru is a curse on India.
    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.
    Sharing Comment via Twitter
    /gnxp/razib-khan-at-the-center-of-eurasia/#comment-1667721
    More... This Commenter This Thread Hide Thread Display All Comments
  2. @Marcus
    Well Bengal is where the Mongolid and Caucasoid traits start to intermix, right? Of course phenotype =/ genotype.
    https://en.wikipedia.org/wiki/Chakma_people
    Coincidentally I recently read something like this http://www.hindustantimes.com/india-news/manipuri-woman-alleges-racism-harassment-at-delhi-airport-s-immigration-desk/story-yoylKApQWGqZRGJwjRNFQM.html

    it’s a pulse admixture happened 1 to 2 thousand years ago. but yeah. also, the south asian ancestry spills over into burma.

    Read More
    • Replies: @Qagan
    Dear Razib Khan,

    Do Southeast Asians such as Burmese, Cambodians, Malays and Thais have minor ANI/West Eurasian ancestry and how much? I ask this because I notice SE Asians usually score quite substantial South Asian component in ADMIXTURE calculators. So I wonder if they have very minor ANI/West Eurasian that comes with the South Asian component or is their "South Asian" component just ASI/ASI-like?

    I will appreciate your generous reply to my inquiry.

    Thank you very much and best regards,
    Qagan

  3. vijay says:
    @Marcus
    Well Bengal is where the Mongolid and Caucasoid traits start to intermix, right? Of course phenotype =/ genotype.
    https://en.wikipedia.org/wiki/Chakma_people
    Coincidentally I recently read something like this http://www.hindustantimes.com/india-news/manipuri-woman-alleges-racism-harassment-at-delhi-airport-s-immigration-desk/story-yoylKApQWGqZRGJwjRNFQM.html

    The entire northeastern population gets abused in Delhi. Correction, all women get abused in Delhi. Correction again, the abuse starts when the train from Assam enters Bihar. Something about women wearing pants, etc. etc. Bimaru is a curse on India.

    Read More
  4. ohwilleke says: • Website

    Like a colossus I look east, and I look west, and stand athwart Eurasia bridging the gap.

    No self-esteem problems for this blogger, no siree! ;)

    About the PCA map.

    * I find it interesting how central the Native American populations are within the Siberian cluster. I would have expected it to be more shifted towards the East Asian side.

    * Query if the cluster from Europe to Siberia to East Asia represents an group of non-Africans sourced in an ancestral Northern route population, while the cluster from Europe to South Asia to Sahul represents a group of non-Africans sourced in an ancestral Southern route population.

    Could the Southern route population would be marginally closer to the African direction of the axis, perhaps because it was a bit earlier than a northern route migration?

    * It is also notable how compact the East Asian population is in PCA phase space compared to the South Asian population, even though both have comparable total population numbers and comparable geographic extent.

    * The key seems to indicate that Southeast Asians are not represented in the data set. Presumably they would be somewhere between South Asians and East Asians.

    * Looks like you could fully reproduce the PCA in quite a bit of detail with admixture blends of four eigenvector populations (Sub-Saharan African, European, East Asian, Sahul).

    * It is also interesting how much of the PCA phase space is completely empty. Tiger Woods would have a spot on the chart a huge distance from every single one of the reference populations somewhere roughly in the middle of both axes.

    Read More
    • Replies: @Shaikorth
    Some things, like the location of the Native Americans in a continuum between modern Europe and East Asia along dimensions (PC’s) 1 and 2 are a feature of the method. They would be more removed from that continuum if further PC’s were shown. In fact, depending on sampling, Native Americans might not be there even along dimensions 1/2 – an example of this happening can be seen in one global PCA included in Lazaridis et al. 2014 supplements. If there were enough Papuans or Native Americans, they would be at the end of PC 2 instead of East Asians, in this analysis they “peak” in other PC’s from 3 onwards.

    PCA and ADMIXTURE are similar in that number of samples, not just their type, has an effect on the result, complicating interpretation. More on that here if you’re interested: http://biorxiv.org/content/early/2016/07/28/066431

    , @Davidski

    I find it interesting how central the Native American populations are within the Siberian cluster. I would have expected it to be more shifted towards the East Asian side.
     
    This is not a typical Global PCA. That's why you can get accurate ancient and recent ancestry proportions from the 10 PCs, even for highly drifted populations.

    Karitiana Indians come out like this.

    Ulchi 64.45
    AfontovaGora3 34.2
    Dai 1.35

    And Kalash...

    Iran_Neolithic:I1945 38.6
    Paniya 20.2
    Yamnaya_Samara:I0357 17.4
    Afanasievo:RISE509 16.55
    Andronovo:RISE505 3.8
    Iran_Late_Neolithic:I1671 3.45
    Iran_Hotu:I1293 0
  5. Shaikorth says:
    @ohwilleke

    Like a colossus I look east, and I look west, and stand athwart Eurasia bridging the gap.
     
    No self-esteem problems for this blogger, no siree! ;)

    About the PCA map.

    * I find it interesting how central the Native American populations are within the Siberian cluster. I would have expected it to be more shifted towards the East Asian side.

    * Query if the cluster from Europe to Siberia to East Asia represents an group of non-Africans sourced in an ancestral Northern route population, while the cluster from Europe to South Asia to Sahul represents a group of non-Africans sourced in an ancestral Southern route population.

    Could the Southern route population would be marginally closer to the African direction of the axis, perhaps because it was a bit earlier than a northern route migration?

    * It is also notable how compact the East Asian population is in PCA phase space compared to the South Asian population, even though both have comparable total population numbers and comparable geographic extent.

    * The key seems to indicate that Southeast Asians are not represented in the data set. Presumably they would be somewhere between South Asians and East Asians.

    * Looks like you could fully reproduce the PCA in quite a bit of detail with admixture blends of four eigenvector populations (Sub-Saharan African, European, East Asian, Sahul).

    * It is also interesting how much of the PCA phase space is completely empty. Tiger Woods would have a spot on the chart a huge distance from every single one of the reference populations somewhere roughly in the middle of both axes.

    Some things, like the location of the Native Americans in a continuum between modern Europe and East Asia along dimensions (PC’s) 1 and 2 are a feature of the method. They would be more removed from that continuum if further PC’s were shown. In fact, depending on sampling, Native Americans might not be there even along dimensions 1/2 – an example of this happening can be seen in one global PCA included in Lazaridis et al. 2014 supplements. If there were enough Papuans or Native Americans, they would be at the end of PC 2 instead of East Asians, in this analysis they “peak” in other PC’s from 3 onwards.

    PCA and ADMIXTURE are similar in that number of samples, not just their type, has an effect on the result, complicating interpretation. More on that here if you’re interested: http://biorxiv.org/content/early/2016/07/28/066431

    Read More
    • Replies: @Davidski

    PCA and ADMIXTURE are similar in that number of samples, not just their type, has an effect on the result, complicating interpretation.
     
    Your post in irrelevant to this PCA.

    Use the 10 PCs and you will get accurate models.
  6. Davidski says:
    @ohwilleke

    Like a colossus I look east, and I look west, and stand athwart Eurasia bridging the gap.
     
    No self-esteem problems for this blogger, no siree! ;)

    About the PCA map.

    * I find it interesting how central the Native American populations are within the Siberian cluster. I would have expected it to be more shifted towards the East Asian side.

    * Query if the cluster from Europe to Siberia to East Asia represents an group of non-Africans sourced in an ancestral Northern route population, while the cluster from Europe to South Asia to Sahul represents a group of non-Africans sourced in an ancestral Southern route population.

    Could the Southern route population would be marginally closer to the African direction of the axis, perhaps because it was a bit earlier than a northern route migration?

    * It is also notable how compact the East Asian population is in PCA phase space compared to the South Asian population, even though both have comparable total population numbers and comparable geographic extent.

    * The key seems to indicate that Southeast Asians are not represented in the data set. Presumably they would be somewhere between South Asians and East Asians.

    * Looks like you could fully reproduce the PCA in quite a bit of detail with admixture blends of four eigenvector populations (Sub-Saharan African, European, East Asian, Sahul).

    * It is also interesting how much of the PCA phase space is completely empty. Tiger Woods would have a spot on the chart a huge distance from every single one of the reference populations somewhere roughly in the middle of both axes.

    I find it interesting how central the Native American populations are within the Siberian cluster. I would have expected it to be more shifted towards the East Asian side.

    This is not a typical Global PCA. That’s why you can get accurate ancient and recent ancestry proportions from the 10 PCs, even for highly drifted populations.

    Karitiana Indians come out like this.

    Ulchi 64.45
    AfontovaGora3 34.2
    Dai 1.35

    And Kalash…

    Iran_Neolithic:I1945 38.6
    Paniya 20.2
    Yamnaya_Samara:I0357 17.4
    Afanasievo:RISE509 16.55
    Andronovo:RISE505 3.8
    Iran_Late_Neolithic:I1671 3.45
    Iran_Hotu:I1293 0

    Read More
  7. Davidski says:
    @Shaikorth
    Some things, like the location of the Native Americans in a continuum between modern Europe and East Asia along dimensions (PC’s) 1 and 2 are a feature of the method. They would be more removed from that continuum if further PC’s were shown. In fact, depending on sampling, Native Americans might not be there even along dimensions 1/2 – an example of this happening can be seen in one global PCA included in Lazaridis et al. 2014 supplements. If there were enough Papuans or Native Americans, they would be at the end of PC 2 instead of East Asians, in this analysis they “peak” in other PC’s from 3 onwards.

    PCA and ADMIXTURE are similar in that number of samples, not just their type, has an effect on the result, complicating interpretation. More on that here if you’re interested: http://biorxiv.org/content/early/2016/07/28/066431

    PCA and ADMIXTURE are similar in that number of samples, not just their type, has an effect on the result, complicating interpretation.

    Your post in irrelevant to this PCA.

    Use the 10 PCs and you will get accurate models.

    Read More
    • Replies: @Shaikorth
    What I said about number of samples affecting the positioning of said samples applies to all PCA's, it isn't a personal criticism. This has been demonstrated in studies:

    http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1000686

    http://journals.plos.org/plosgenetics/article/figure/image?id=info%3Adoi/10.1371/journal.pgen.1000686.g003&size=large

    Coincidence or not, C resembles the common "V" shape of a global PCA done on human populations.

    If the sample sizes are not even there will be a shift compared to a PCA that has even sampling.
    But it doesn't necessarily mean a model using multiple dimensions produces "wrong" results even if sample sizes are "uneven". I don't think your Karitiana result is really off the mark.

  8. Shaikorth says:
    @Davidski

    PCA and ADMIXTURE are similar in that number of samples, not just their type, has an effect on the result, complicating interpretation.
     
    Your post in irrelevant to this PCA.

    Use the 10 PCs and you will get accurate models.

    What I said about number of samples affecting the positioning of said samples applies to all PCA’s, it isn’t a personal criticism. This has been demonstrated in studies:

    http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1000686

    http://journals.plos.org/plosgenetics/article/figure/image?id=info%3Adoi/10.1371/journal.pgen.1000686.g003&size=large

    Coincidence or not, C resembles the common “V” shape of a global PCA done on human populations.

    If the sample sizes are not even there will be a shift compared to a PCA that has even sampling.
    But it doesn’t necessarily mean a model using multiple dimensions produces “wrong” results even if sample sizes are “uneven”. I don’t think your Karitiana result is really off the mark.

    Read More
    • Replies: @Davidski

    What I said about number of samples affecting the positioning of said samples applies to all PCA’s, it isn’t a personal criticism.
     
    Once you pass critical mass nothing changes.

    The challenge with PCA is not to let highly drifted populations hijack the analysis.

    I don’t think your Karitiana result is really off the mark.
     
    What makes you think it's off the mark at all?

    Karitiana are modeled as East Asian, MA1 and Onge in latest scientific literature.

    That's essentially the same thing as my model, except I'm using the more relevant AfontovaGora3 and probably the more relevant Ulchi.

    Dai sits better with these two, although Onge works too. Let's see what happens when latest scientific literature starts using AfontovaGora3 and close relatives, instead of MA1.
  9. Davidski says:
    @Shaikorth
    What I said about number of samples affecting the positioning of said samples applies to all PCA's, it isn't a personal criticism. This has been demonstrated in studies:

    http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1000686

    http://journals.plos.org/plosgenetics/article/figure/image?id=info%3Adoi/10.1371/journal.pgen.1000686.g003&size=large

    Coincidence or not, C resembles the common "V" shape of a global PCA done on human populations.

    If the sample sizes are not even there will be a shift compared to a PCA that has even sampling.
    But it doesn't necessarily mean a model using multiple dimensions produces "wrong" results even if sample sizes are "uneven". I don't think your Karitiana result is really off the mark.

    What I said about number of samples affecting the positioning of said samples applies to all PCA’s, it isn’t a personal criticism.

    Once you pass critical mass nothing changes.

    The challenge with PCA is not to let highly drifted populations hijack the analysis.

    I don’t think your Karitiana result is really off the mark.

    What makes you think it’s off the mark at all?

    Karitiana are modeled as East Asian, MA1 and Onge in latest scientific literature.

    That’s essentially the same thing as my model, except I’m using the more relevant AfontovaGora3 and probably the more relevant Ulchi.

    Dai sits better with these two, although Onge works too. Let’s see what happens when latest scientific literature starts using AfontovaGora3 and close relatives, instead of MA1.

    Read More
    • Replies: @Shaikorth


    Once you pass critical mass nothing changes.

    The challenge with PCA is not to let highly drifted populations hijack the analysis.
     

    There probably is always going to be a drift dimension at some point even if it's not among the most important ones (for example, one defined by modern Natives, not East Asians/Onge or ANE which are their likely ancestral populations). Even the East Asian dimension could be a "drift dimension" if Lazaridis' "Onge-ANE"-type models turns out correct. But using multiple dimensions quite demonstrably produces models that look fitting. My reply to ohwilleke was to note just two dimensions can give odd impressions (like Native Americans as modern European + East Asian).

    What makes you think it’s off the mark at all?

    Karitiana are modeled as East Asian, MA1 and Onge in latest scientific literature.

    That’s essentially the same thing as my model, except I’m using the more relevant AfontovaGora3 and probably the more relevant Ulchi.

     

    We don't have the actual Beringian mixing populations yet so "not really off the mark" should be viewed as a compliment in this case. It's probably how the scientists working on this stuff consider their present models, though I don't think there's going to be much change with more relevant samples. They've been occasionally using AG2 and EHG instead of MA-1 as ANE but the results aren't wildly different. Since we're on this issue, does this show Onge-like stuff Skoglund & co found in Karitiana if you use AG3, Dai and Onge as sources?
  10. Shaikorth says:
    @Davidski

    What I said about number of samples affecting the positioning of said samples applies to all PCA’s, it isn’t a personal criticism.
     
    Once you pass critical mass nothing changes.

    The challenge with PCA is not to let highly drifted populations hijack the analysis.

    I don’t think your Karitiana result is really off the mark.
     
    What makes you think it's off the mark at all?

    Karitiana are modeled as East Asian, MA1 and Onge in latest scientific literature.

    That's essentially the same thing as my model, except I'm using the more relevant AfontovaGora3 and probably the more relevant Ulchi.

    Dai sits better with these two, although Onge works too. Let's see what happens when latest scientific literature starts using AfontovaGora3 and close relatives, instead of MA1.

    Once you pass critical mass nothing changes.

    The challenge with PCA is not to let highly drifted populations hijack the analysis.

    There probably is always going to be a drift dimension at some point even if it’s not among the most important ones (for example, one defined by modern Natives, not East Asians/Onge or ANE which are their likely ancestral populations). Even the East Asian dimension could be a “drift dimension” if Lazaridis’ “Onge-ANE”-type models turns out correct. But using multiple dimensions quite demonstrably produces models that look fitting. My reply to ohwilleke was to note just two dimensions can give odd impressions (like Native Americans as modern European + East Asian).

    What makes you think it’s off the mark at all?

    Karitiana are modeled as East Asian, MA1 and Onge in latest scientific literature.

    That’s essentially the same thing as my model, except I’m using the more relevant AfontovaGora3 and probably the more relevant Ulchi.

    We don’t have the actual Beringian mixing populations yet so “not really off the mark” should be viewed as a compliment in this case. It’s probably how the scientists working on this stuff consider their present models, though I don’t think there’s going to be much change with more relevant samples. They’ve been occasionally using AG2 and EHG instead of MA-1 as ANE but the results aren’t wildly different. Since we’re on this issue, does this show Onge-like stuff Skoglund & co found in Karitiana if you use AG3, Dai and Onge as sources?

    Read More
    • Replies: @Davidski
    Correction: Onge sits better than Dai. Onge disappears when Ulchi is dropped, but the fit is horrible.

    Karitiana
    Ulchi 64.65
    AfontovaGora3 33.3
    Andamanese_Onge 1.9
    Dai 0.15

    Karitiana
    Ulchi 64.7
    AfontovaGora3 33.25
    Andamanese_Onge 2.05

    Karitiana
    AfontovaGora3 58.2
    Dai 41.8
    Andamanese_Onge 0
  11. Davidski says:
    @Shaikorth


    Once you pass critical mass nothing changes.

    The challenge with PCA is not to let highly drifted populations hijack the analysis.
     

    There probably is always going to be a drift dimension at some point even if it's not among the most important ones (for example, one defined by modern Natives, not East Asians/Onge or ANE which are their likely ancestral populations). Even the East Asian dimension could be a "drift dimension" if Lazaridis' "Onge-ANE"-type models turns out correct. But using multiple dimensions quite demonstrably produces models that look fitting. My reply to ohwilleke was to note just two dimensions can give odd impressions (like Native Americans as modern European + East Asian).

    What makes you think it’s off the mark at all?

    Karitiana are modeled as East Asian, MA1 and Onge in latest scientific literature.

    That’s essentially the same thing as my model, except I’m using the more relevant AfontovaGora3 and probably the more relevant Ulchi.

     

    We don't have the actual Beringian mixing populations yet so "not really off the mark" should be viewed as a compliment in this case. It's probably how the scientists working on this stuff consider their present models, though I don't think there's going to be much change with more relevant samples. They've been occasionally using AG2 and EHG instead of MA-1 as ANE but the results aren't wildly different. Since we're on this issue, does this show Onge-like stuff Skoglund & co found in Karitiana if you use AG3, Dai and Onge as sources?

    Correction: Onge sits better than Dai. Onge disappears when Ulchi is dropped, but the fit is horrible.

    Karitiana
    Ulchi 64.65
    AfontovaGora3 33.3
    Andamanese_Onge 1.9
    Dai 0.15

    Karitiana
    Ulchi 64.7
    AfontovaGora3 33.25
    Andamanese_Onge 2.05

    Karitiana
    AfontovaGora3 58.2
    Dai 41.8
    Andamanese_Onge 0

    Read More
    • Replies: @Shaikorth
    That's interesting. Assuming Ulchi has the amount of ANE and Onge/Dai that the papers propose (using MA-1, AG2 and EHG), the last fit indicates more ANE than the first two.

    Does qpAdm for Karitiana and Ulchi as mixes of AG3 & Dai or Onge work using Mbuti, Ust_Ishim, Papuan, Kostenki14 and Bichon as outgroups?
  12. Razib,

    If you are inclined to do so, you should post your coordinates for the ten dimensions.

    I’ve been working on a modelling setup that works for all of Eurasia, and will post the results at Eurogenes (and some other places), this weekend/early next week. I think you’ll be pleased with what you see. So far, people from peninsular South Asia are getting excellent models, the ASI proportions make total sense.

    Although, it is totally fine if you don’t do so.

    Read More
  13. Shaikorth says:
    @Davidski
    Correction: Onge sits better than Dai. Onge disappears when Ulchi is dropped, but the fit is horrible.

    Karitiana
    Ulchi 64.65
    AfontovaGora3 33.3
    Andamanese_Onge 1.9
    Dai 0.15

    Karitiana
    Ulchi 64.7
    AfontovaGora3 33.25
    Andamanese_Onge 2.05

    Karitiana
    AfontovaGora3 58.2
    Dai 41.8
    Andamanese_Onge 0

    That’s interesting. Assuming Ulchi has the amount of ANE and Onge/Dai that the papers propose (using MA-1, AG2 and EHG), the last fit indicates more ANE than the first two.

    Does qpAdm for Karitiana and Ulchi as mixes of AG3 & Dai or Onge work using Mbuti, Ust_Ishim, Papuan, Kostenki14 and Bichon as outgroups?

    Read More
  14. Halvorson says:

    A French guy on the Anthrogenica forum has improved on the results of Eurogenes’ already great K15 analysis with this mapping tool:

    http://gen3553.pagesperso-orange.fr/ADN/K15.htm

    It would be nice if he could write software that would place the user’s square automatically, but all the same these results are the most accurate I’ve gotten. They placed me exactly on the correct blue square.

    Read More
  15. Qagan says:
    @Razib Khan
    it's a pulse admixture happened 1 to 2 thousand years ago. but yeah. also, the south asian ancestry spills over into burma.

    Dear Razib Khan,

    Do Southeast Asians such as Burmese, Cambodians, Malays and Thais have minor ANI/West Eurasian ancestry and how much? I ask this because I notice SE Asians usually score quite substantial South Asian component in ADMIXTURE calculators. So I wonder if they have very minor ANI/West Eurasian that comes with the South Asian component or is their “South Asian” component just ASI/ASI-like?

    I will appreciate your generous reply to my inquiry.

    Thank you very much and best regards,
    Qagan

    Read More
  16. My Malaysian sister-in-law claims Malaysians are Aryan. That they came over the sea from India to Sumatra and jave and then to Malaysia. But then Malaysian racial politics is a minefield.

    Read More

Comments are closed.

Subscribe to All Razib Khan Comments via RSS
PastClassics
A simple remedy for income stagnation
Confederate Flag Day, State Capitol, Raleigh, N.C. -- March 3, 2007
The major media overlooked Communist spies and Madoff’s fraud. What are they missing today?
Are elite university admissions based on meritocracy and diversity as claimed?
The evidence is clear — but often ignored