The Unz Review - Mobile
A Collection of Interesting, Important, and Controversial Perspectives Largely Excluded from the American Mainstream Media
 BlogviewJames Thompson Archive
The Genetics of University Entrance
🔊 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

Cambridge undergrads

Newspaper reports are still discussing the story about the numbers of Africans admitted to Oxbridge, but I have not seen any giving the numbers of AAA students available, or any that mention intelligence. I doubt that Admissions Officers read my blog, or indeed that they would survive in post if they were ever caught doing so, but it is perfectly possible that Admissions Officers have friends, and some of those friends might innocently print out this post for them to read when they call round for coffee.

First, let us start with some data on cognitive assessments carried out in schools. They are not face to face Wechsler tests, nor are they the traditional tests on which we have decades of comparative data, but they give us some idea of student capabilities by ethnic group, and that is what we need in order to contribute to this debate. Indeed, looking at the data has a generality beyond the specific case of Oxbridge, and applies to all university entrance matters where ethnicity is an issue. Here are cognitive scores for ethnic groups in Britain, as assessed by the CAT test at school.

pupil background CAT

In the table below I will concentrate mostly on those groups with more than 1000 subjects and not having “other” in their ethnicity description, since one does not know what ethnicities are included. I simplify the three reasoning measures into one simple average and standard deviation (this is crude but quick). I then give the percentage of each ethnic group who are above IQ 130, on the basis of a requirement that students are IQ 130 and above on the assumption that the population has an average IQ 100 and a standard deviation of 15 points. These are all approximations, since the observed standard deviations are narrower than expected, possibly because the tests may not have been offered to some lower ability children, and they may have omitted private schools with higher ability students. I have rounded the means and standard deviation scores up or down with the usual 0.5 break point, which will have further affected some of the calculations.

130+ by ethnicity

As you can see, on these figures Chinese students are the most likely to get to a good university. They would do so at roughly 3 times the rate of white students. White students would be almost 7 times more likely to get to these universities than Black Caribbean students.

Remember also that category of “3 A’s or better” hides a four point scale, ranging from AAA to A*A*A*, so a demanding university can still pick and choose within the students who achieve the minimal entrance level.

Now back to the actual figures of students getting AAA, which is what the universities must deal with. Since only 6600 applicants get a place at Oxbridge, we can assume that if every candidate with at least 3 As applies, then the success rate is 6600/17,146 which is 38%. The final column shows what the actual figures of placement offers should be for Oxbridge. These would be the admissions made on merit only, as judged by scholastic results. There would only be 1 Black Caribbean for every other 277 students.

Oxbridge entry

I do not frequent these establishments, but if you have a pressing interest in using genetic background as an entry criterion, a prospect which does not please me, you may wish to go around counting undergraduates and rate them according to their apparent ethnicities, and compare them with the various calculation shown above. The final point should be very clear: entrance to university cannot be calculated on population numbers alone.

 
• Category: Science • Tags: Academia, IQ 
Hide 68 CommentsLeave a Comment
Commenters to Ignore...to FollowEndorsed Only
Trim Comments?
    []
  1. dearieme says:

    There is another way to find evidence that the dons aren’t a bunch of Literally Hitlers. Look at the racial mix of the postgraduate admissions. Postgraduate admission is less competitive while still being attractive to clever students from around the world.

    Just peek into a College Hall at lunchtime outside full term. The undergraduates are “down” and the graduate students are still “up”. At a casual glance I fail to see any evidence at all of carpet-biting Nazism. Even the blind could guess at the diversity: the noise can only mean that there are plenty of Chinese and Americans present.

    Read More
    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 Display All Comments
  2. Made an under-estimate of what is covered in “better than AAA”. It actually goes to A*A*A*A*A* plus a Subsiduary A. Oxbridge can pick and choose. Not strictly relevant here, but the quality varies by college.

    Read More
    • Replies: @AnotherDad
    I simply don't know enough about the British testing regime. (This CAT test doesn't even show up in Wikipedia--for CAT i get the common admissions test in India.)

    But obvious question--how comprehensive. The UK has--ballpark--a quarter the population of the US. Our (American) cohorts run 4m per year. That puts Britain at 1m--a quick search suggests that's pretty spot on. So why is this at a measly 160K? Which 160K is it?

    ~~

    Ok, a bit more poking i found there is something called the Classics Admissions Test, for which scores are required for admission to Oxford. I assume this is it.

    But this will be very self-selected--which explains some of the otherwise incomprehensible numbers (like decent mixed white-black score).

    Tacking on a bit of explanation--what the test is, who takes it--to the post would help.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  3. dearieme says:
    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  4. dearieme says:
    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  5. res says:

    Thank you for the analysis. I don’t know if it is just me, but I found your last graphic a bit hard to understand at first glance. Some things which might help:
    - Be more explicit about what “Russell” is. I interpret it as the number of people in each group who achieve AAA or better. But a text search shows no “Russell” anywhere else.
    - Move the percentage column to the far right to make it more obviously linked with Oxbridge admissions (further away from Russell column). Maybe also add a column heading.
    - Add the 6600 total admittees to the bottom row. Since that is the denominator for your percentages I think it would help make their meaning more apparent.

    I also think it would be helpful to show the total population percentages for each group to emphasize the point of how different those might be. It is a bit of a duplication from your second graphic, but I think it would be worth it. Maybe even show the ratio between those two percentages.

    P.S. Feel free not to pass this comment. I just thought it might be good to strive for clarity since I think you are going for a less technical audience with this post.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  6. dux.ie says:

    Just ask the newspapers where they want to place Oxbridge with respect to the 1339 US universities and colleges. There is a very very wide range for them to choose. Plenty of room on the left hand side.

    http://tinypic.com/view.php?pic=s6nf9t&s=9

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  7. Agree with all your points! I need a graphic artist, and used to hire one occasionally. Russell is short-hand for the Russell group of universities.

    The Russell Group was formed in 1994 by 17 British research universities – Birmingham, Bristol, Cambridge, Edinburgh, Glasgow, Imperial College London, Queen Mary, Leeds, Liverpool, London School of Economics, Manchester, Newcastle, Nottingham, Oxford, Sheffield, Southampton, University College London and Warwick.

    “Oxbridge” is a small sub-set.

    I was going to do the whole thing again, but then felt it was duplicating things already said, and would delay things too much.

    Read More
    • Replies: @res
    Thanks for your response! So to help me understand better, Russell is this: https://en.wikipedia.org/wiki/Russell_Group
    Right?

    If I understand correctly, you are using Russell admissions as a proxy for "AAA students." Is that right? How good a proxy is that? In the US affirmative action (present at essentially all universities) would make similar proxies very imprecise. Are A-level grade statistics by race available?

    The data and graphic here provide a great look at A-level grade inflation:
    https://en.wikipedia.org/wiki/GCE_Advanced_Level_(United_Kingdom)#Awarding
    The rest of the article includes some good background for the non-Brits (like me) here.

    Here are statistics for AAA or better by race: https://www.ethnicity-facts-figures.service.gov.uk/education-skills-and-training/a-levels/draft-percentage-of-students-achieving-3-a-grades-or-better-at-a-level/latest
    The statistics for 2016 look like the numbers you are using so I guess I was wrong about the proxy idea. They also include the percentages for each race to allow computing the overall test taking populations.

    Now that I understand the data better, I think a good metric to look at is the ratio of the Russell and Oxbridge columns. Perhaps best expressed as Oxbridge/Russell which I interpret as the percentage of "eligible" (AAA or better) students from each group who are admitted to Oxbridge. A quick eyeball of those numbers (all around 40%) looks surprisingly equitable. If equivalent statistics were generated for the rough equivalent of HYP (Harvard Yale Princeton) I would expect affirmative action to play a much larger role.

    Contrasting those numbers with the overall population percentages of each group would help clarify how your perspective relates to the overall population representation appearance and make clear how well your AAA idea explains the non-discriminatory reality. I think the overall population percentages could be calculated by adding the race AAA percentages as follows:
    Russell / (race AAA %) = total group test taking population
    Oxbridge / (total group test taking population) = % of each group admitted to Oxbridge

    That is for the test taking populations, which I think is much more relevant than the all ages populations.

    Does this match your sense of what the data is indicating?

    P.S. I would be happy to lend a hand with data analysis and interpretation (and basic graphic creation), but I do not have the visual design sense of a graphic artist.
    , @Philip Owen
    That's 18 and doesn't include Cardiff FF which was one of the first 12.

    24 nowmanyway. Not quite all the Redbricks.

    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  8. DFH says:

    I assume that IQ differences also explain most of the difference in admissions between state/private and uppermiddle/working class

    Read More
    • Replies: @gate666
    what is the reason for discrepancy between lynns estimates and these.
    , @Edward
    It would be most prudent of us to think so, yes.

    The CAT data to which Dr. Thompson refers found that those attending independent (private) primary schools have an average IQ of 111.

    Similarly, children entitled to a free school meal had an average IQ of 93, while those who were not entitled to these meals had an average IQ of 102.

    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  9. res says:
    @James Thompson
    Agree with all your points! I need a graphic artist, and used to hire one occasionally. Russell is short-hand for the Russell group of universities.

    The Russell Group was formed in 1994 by 17 British research universities – Birmingham, Bristol, Cambridge, Edinburgh, Glasgow, Imperial College London, Queen Mary, Leeds, Liverpool, London School of Economics, Manchester, Newcastle, Nottingham, Oxford, Sheffield, Southampton, University College London and Warwick.

    "Oxbridge" is a small sub-set.

    I was going to do the whole thing again, but then felt it was duplicating things already said, and would delay things too much.

    Thanks for your response! So to help me understand better, Russell is this: https://en.wikipedia.org/wiki/Russell_Group
    Right?

    If I understand correctly, you are using Russell admissions as a proxy for “AAA students.” Is that right? How good a proxy is that? In the US affirmative action (present at essentially all universities) would make similar proxies very imprecise. Are A-level grade statistics by race available?

    The data and graphic here provide a great look at A-level grade inflation:

    https://en.wikipedia.org/wiki/GCE_Advanced_Level_(United_Kingdom)#Awarding

    The rest of the article includes some good background for the non-Brits (like me) here.

    Here are statistics for AAA or better by race: https://www.ethnicity-facts-figures.service.gov.uk/education-skills-and-training/a-levels/draft-percentage-of-students-achieving-3-a-grades-or-better-at-a-level/latest
    The statistics for 2016 look like the numbers you are using so I guess I was wrong about the proxy idea. They also include the percentages for each race to allow computing the overall test taking populations.

    Now that I understand the data better, I think a good metric to look at is the ratio of the Russell and Oxbridge columns. Perhaps best expressed as Oxbridge/Russell which I interpret as the percentage of “eligible” (AAA or better) students from each group who are admitted to Oxbridge. A quick eyeball of those numbers (all around 40%) looks surprisingly equitable. If equivalent statistics were generated for the rough equivalent of HYP (Harvard Yale Princeton) I would expect affirmative action to play a much larger role.

    Contrasting those numbers with the overall population percentages of each group would help clarify how your perspective relates to the overall population representation appearance and make clear how well your AAA idea explains the non-discriminatory reality. I think the overall population percentages could be calculated by adding the race AAA percentages as follows:
    Russell / (race AAA %) = total group test taking population
    Oxbridge / (total group test taking population) = % of each group admitted to Oxbridge

    That is for the test taking populations, which I think is much more relevant than the all ages populations.

    Does this match your sense of what the data is indicating?

    P.S. I would be happy to lend a hand with data analysis and interpretation (and basic graphic creation), but I do not have the visual design sense of a graphic artist.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  10. gate666 says:
    @DFH
    I assume that IQ differences also explain most of the difference in admissions between state/private and uppermiddle/working class

    what is the reason for discrepancy between lynns estimates and these.

    Read More
    • Replies: @James Thompson
    Probably different samples, and need also to see how CAT performs viz a viz Raven's Matrices, for example. However, the discrepancies should not be very large, and the actual sample sizes here are very big, so unless there are real problems with CAT as a measure, this would be a good data set for scholastic predictions. (Remember also that all the standard deviations are rather narrow, so that needs some explaining).
    , @Edward
    To which estimates are you referring? These data are much more recent than most of Lynn's estimates, and have a substantially larger sample size (10 times larger, or more, than most of Lynn's studies).
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  11. @gate666
    what is the reason for discrepancy between lynns estimates and these.

    Probably different samples, and need also to see how CAT performs viz a viz Raven’s Matrices, for example. However, the discrepancies should not be very large, and the actual sample sizes here are very big, so unless there are real problems with CAT as a measure, this would be a good data set for scholastic predictions. (Remember also that all the standard deviations are rather narrow, so that needs some explaining).

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  12. Edward says:
    @DFH
    I assume that IQ differences also explain most of the difference in admissions between state/private and uppermiddle/working class

    It would be most prudent of us to think so, yes.

    The CAT data to which Dr. Thompson refers found that those attending independent (private) primary schools have an average IQ of 111.

    Similarly, children entitled to a free school meal had an average IQ of 93, while those who were not entitled to these meals had an average IQ of 102.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  13. Edward says:
    @gate666
    what is the reason for discrepancy between lynns estimates and these.

    To which estimates are you referring? These data are much more recent than most of Lynn’s estimates, and have a substantially larger sample size (10 times larger, or more, than most of Lynn’s studies).

    Read More
    • Replies: @gate666
    lynn claimed iq of british indians was around 83.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  14. @James Thompson
    Agree with all your points! I need a graphic artist, and used to hire one occasionally. Russell is short-hand for the Russell group of universities.

    The Russell Group was formed in 1994 by 17 British research universities – Birmingham, Bristol, Cambridge, Edinburgh, Glasgow, Imperial College London, Queen Mary, Leeds, Liverpool, London School of Economics, Manchester, Newcastle, Nottingham, Oxford, Sheffield, Southampton, University College London and Warwick.

    "Oxbridge" is a small sub-set.

    I was going to do the whole thing again, but then felt it was duplicating things already said, and would delay things too much.

    That’s 18 and doesn’t include Cardiff FF which was one of the first 12.

    24 nowmanyway. Not quite all the Redbricks.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  15. This reminds me of a time when I went to study for a national exam for my profession at the main library on the University of Texas campus in Austin. It was early on a Saturday morning and I was there before they opened the doors to the library. I looked around and noticed the ethnic mix of the students waiting to get in and study on a Saturday morning. They were about half white and half Asian. I was concerned that the University was excluding blacks and Hispanics. However, when I took a break for lunch in a nearby campus mall, there were plenty of blacks and Hispanics hanging out, playing video games and pool, etc. Interesting.

    Read More
    • Replies: @lavoisier

    This reminds me of a time when I went to study for a national exam for my profession at the main library on the University of Texas campus in Austin. It was early on a Saturday morning and I was there before they opened the doors to the library. I looked around and noticed the ethnic mix of the students waiting to get in and study on a Saturday morning. They were about half white and half Asian. I was concerned that the University was excluding blacks and Hispanics. However, when I took a break for lunch in a nearby campus mall, there were plenty of blacks and Hispanics hanging out, playing video games and pool, etc. Interesting.
     
    Didn't you know that the library on Saturday is closed to black and Hispanic students?
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  16. anon[508] • Disclaimer says:

    We should just make all our elite universities 100% black: Oxbridge, LSE, Ivy League, MIT, Stanford. You can name any reason and it’d be legit — reparation, boosting their self-esteem, preventing them from having to study alongside whites and asians which would lower their self-esteem and performance due to racism from those groups etc.

    While we’re at it also make sure the faculty and administration are 100% black. No whites allowed. Do this until black and white income are at parity.

    Read More
    • Replies: @dearieme
    Interesting idea. I propose a careful trial - begin with the LSE. There would be so little to lose.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  17. dearieme says:
    @anon
    We should just make all our elite universities 100% black: Oxbridge, LSE, Ivy League, MIT, Stanford. You can name any reason and it'd be legit -- reparation, boosting their self-esteem, preventing them from having to study alongside whites and asians which would lower their self-esteem and performance due to racism from those groups etc.

    While we're at it also make sure the faculty and administration are 100% black. No whites allowed. Do this until black and white income are at parity.

    Interesting idea. I propose a careful trial – begin with the LSE. There would be so little to lose.

    Read More
    • LOL: Philip Owen
    • Replies: @jim jones
    Black MPs in the UK are always complaining about the lack of ethnic students accepted by Oxbridge:

    https://www.theguardian.com/education/2017/oct/19/oxford-accused-of-social-apartheid-as-colleges-admit-no-black-students
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  18. lavoisier says: • Website
    @Hannah Katz
    This reminds me of a time when I went to study for a national exam for my profession at the main library on the University of Texas campus in Austin. It was early on a Saturday morning and I was there before they opened the doors to the library. I looked around and noticed the ethnic mix of the students waiting to get in and study on a Saturday morning. They were about half white and half Asian. I was concerned that the University was excluding blacks and Hispanics. However, when I took a break for lunch in a nearby campus mall, there were plenty of blacks and Hispanics hanging out, playing video games and pool, etc. Interesting.

    This reminds me of a time when I went to study for a national exam for my profession at the main library on the University of Texas campus in Austin. It was early on a Saturday morning and I was there before they opened the doors to the library. I looked around and noticed the ethnic mix of the students waiting to get in and study on a Saturday morning. They were about half white and half Asian. I was concerned that the University was excluding blacks and Hispanics. However, when I took a break for lunch in a nearby campus mall, there were plenty of blacks and Hispanics hanging out, playing video games and pool, etc. Interesting.

    Didn’t you know that the library on Saturday is closed to black and Hispanic students?

    Read More
    • Replies: @Jim Bob Lassiter
    Didn't you know that "black" should be "Black" no matter where it is found? Do that again and I'm gonna send you up front to Hermana Shaniqua for a good ol' fashion hard knuckle rappin' with a steel ruler.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  19. I wonder if anyone can offer any insight on this subject matter as it apples to university admissions practices in Mexico.

    https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/?utm_source=Mexico+News+Daily&utm_campaign=197a5a9f33-newsletter_COPY_02&utm_medium=email&utm_term=0_f1536a3787-197a5a9f33-349530129

    Fred Reed?? Allan Wall?? José Vasconcelos?? Jorge Ramos?? Esther Gordillo??

    Read More
    • Replies: @dux.ie
    Part of the attribute of people with university degree is signalling that they are a cut above the average. If more people get university degrees then that might not worth the papers that certified them. In USA the NewYorkFed data showed that for some studies the percentages of university graduates working in jobs that do not require degrees (under-employment) are very high, for example CriminalJustice 75.5%, PerformingArts 65.5%, LiberalArts 57%, MassMedia 56.1%, Sociology 52.1%, EthnicStudies 50.5%, etc. And they might already have acquired hefty student loans to achieve that.

    Another is the international competitiveness of the country's graduates. The nominal IQ of university graduate is 115, but many countries trying to cut corners and devalueing their worth. The international competitiveness of the graduate is dependent on the average national IQ and the fraction of the population with degrees which I measured with ENIGMA (Estimated Nominal IQ of Graduate from Multiple Attributes). If the average national IQ is low then the country should reduce the fraction of the population with degrees, otherwise their degrees are less worthy internationally. Example,

    Rank ENIGMA IQLynn UniA Frac115 SRatio Country
    2 118.04 102 0.142 0.193 1.36 ITA
    3 117.60 100 0.120 0.159 1.32 AUT
    4 114.84 106 0.278 0.274 0.99 KOR
    5 114.51 105 0.263 0.252 0.96 JPN
    6 114.49 99 0.151 0.143 0.95 DEU
    24 108.47 100 0.286 0.159 0.55 GBR
    25 108.29 99 0.268 0.143 0.53 CAN
    33 105.54 98 0.308 0.129 0.42 USA
    38 102.78 88 0.162 0.036 0.22 MEX

    The national average IQ data from Lynn for Mexico is 88. From OECD data the fraction of Mexican with degrees is 0.162, that gives ENIGMA=102.8 and signalling ratio SRatio=0.22, already way below the nominal ENIGMA value of 115.

    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162. Dispite what that person interviewed said, the prestige of that university might already below average withn the country, unless it is in a very high IQ region. To reach ENIGMA=115 that fraction should be 0.036
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  20. @lavoisier

    This reminds me of a time when I went to study for a national exam for my profession at the main library on the University of Texas campus in Austin. It was early on a Saturday morning and I was there before they opened the doors to the library. I looked around and noticed the ethnic mix of the students waiting to get in and study on a Saturday morning. They were about half white and half Asian. I was concerned that the University was excluding blacks and Hispanics. However, when I took a break for lunch in a nearby campus mall, there were plenty of blacks and Hispanics hanging out, playing video games and pool, etc. Interesting.
     
    Didn't you know that the library on Saturday is closed to black and Hispanic students?

    Didn’t you know that “black” should be “Black” no matter where it is found? Do that again and I’m gonna send you up front to Hermana Shaniqua for a good ol’ fashion hard knuckle rappin’ with a steel ruler.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  21. dux.ie says:
    @Jim Bob Lassiter
    I wonder if anyone can offer any insight on this subject matter as it apples to university admissions practices in Mexico.

    https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/?utm_source=Mexico+News+Daily&utm_campaign=197a5a9f33-newsletter_COPY_02&utm_medium=email&utm_term=0_f1536a3787-197a5a9f33-349530129

    Fred Reed?? Allan Wall?? José Vasconcelos?? Jorge Ramos?? Esther Gordillo??

    Part of the attribute of people with university degree is signalling that they are a cut above the average. If more people get university degrees then that might not worth the papers that certified them. In USA the NewYorkFed data showed that for some studies the percentages of university graduates working in jobs that do not require degrees (under-employment) are very high, for example CriminalJustice 75.5%, PerformingArts 65.5%, LiberalArts 57%, MassMedia 56.1%, Sociology 52.1%, EthnicStudies 50.5%, etc. And they might already have acquired hefty student loans to achieve that.

    Another is the international competitiveness of the country’s graduates. The nominal IQ of university graduate is 115, but many countries trying to cut corners and devalueing their worth. The international competitiveness of the graduate is dependent on the average national IQ and the fraction of the population with degrees which I measured with ENIGMA (Estimated Nominal IQ of Graduate from Multiple Attributes). If the average national IQ is low then the country should reduce the fraction of the population with degrees, otherwise their degrees are less worthy internationally. Example,

    Rank ENIGMA IQLynn UniA Frac115 SRatio Country
    2 118.04 102 0.142 0.193 1.36 ITA
    3 117.60 100 0.120 0.159 1.32 AUT
    4 114.84 106 0.278 0.274 0.99 KOR
    5 114.51 105 0.263 0.252 0.96 JPN
    6 114.49 99 0.151 0.143 0.95 DEU
    24 108.47 100 0.286 0.159 0.55 GBR
    25 108.29 99 0.268 0.143 0.53 CAN
    33 105.54 98 0.308 0.129 0.42 USA
    38 102.78 88 0.162 0.036 0.22 MEX

    The national average IQ data from Lynn for Mexico is 88. From OECD data the fraction of Mexican with degrees is 0.162, that gives ENIGMA=102.8 and signalling ratio SRatio=0.22, already way below the nominal ENIGMA value of 115.

    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162. Dispite what that person interviewed said, the prestige of that university might already below average withn the country, unless it is in a very high IQ region. To reach ENIGMA=115 that fraction should be 0.036

    Read More
    • Replies: @res
    I like your ENIGMA methodology. Do you expand on it anywhere (e.g. the full table of calculations)? The only thing I was able to find was this very interesting comment (and followups) at West Hunter: https://westhunt.wordpress.com/2017/06/17/african-variation/#comment-92877

    Have you tried cross checking your Frac115 numbers with PISA results?

    Could you elaborate a bit more on which OECD data (e.g. year, age range, variables included in UniA) you used and which source had the data with three significant digits? The best I am finding is Table A1.1 and A1.2 of Education at a Glance: http://www.oecd.org/education/education-at-a-glance-19991487.htm
    Annex 3 has more detail including a link to the spreadsheet below: www.oecd.org/edu/skills-beyond-school/EAG2017-Annex-3.pdf
    Here are the 2016 Tables in Excel: http://dx.doi.org/10.1787/888933562885

    I assume ENIGMA is just the UniA converted to an IQ delta (Z-score * 15) and added to IQLynn?

    The big issue I see is that I think you are using the Table A1.1 data (ages 25-64) while Table A1.2 (the spreadsheet version is much better than the PDF because it breaks out the different levels of tertiary education) shows that for a country like Mexico UniA varies by about a factor of two between the 25-34 and 55-64 age groups. That effect is probably partially canceled by the Flynn Effect, but I am not sure how to correct for those.


    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162.
     
    I think you are comparing apples and oranges here. 92k is the number who wrote the entrance exam. Your other statistic is based on total population. What is the percentage of the student age population who applied to the university? The data I see in Table A1.1 for Mexico (exclusive categories) is 19% complete secondary and 15% have a bachelors or above. So even if everyone who completes secondary (~34%) applies you need to divide your 0.251 number by 3.

    P.S. The data tables in the PDF document do not include a description of the data annotations, so here they are from the spreadsheet.

    Symbols for missing data

    m Data are not available
    a Data are not applicable because the category does not apply
    x Data included in another category
    d Includes data from another category
    0 Magnitude is either negligible or zero
    c There are too few observations to provide reliable estimates.
    r Values are below a certain reliability threshold and should be interpreted with caution
    b Break in the time series

    , @Kratoklastes

    The nominal IQ of university graduate is 115
     
    That aggregate statistic is pretty pointless, since it conceals significant variation by institution and discipline: the median graduate in Physics from Cal Tech or MIT is a different species from the median graduate in Gender Studies from Pomona (and Pomona's not a bad school).

    Go further down the school pile, and push the discipline further towards "hobby" status, and watch the median student's IQ drop toward the mean.

    In other words, there are significantly-different sub-populations within the 'university graduate' cohort: for some of those sub-populations, 115 is a significant under-estimate.

    And let's stipulate: the reason why a shitload of graduates work in areas that do not require a degree, is because anyone below the top decile of graduates is not useful in their discipline.

    Nobody outside of the top 5% of the graduating class is getting a discipline-specific job straight out of their undergraduate degree (with teensy tiny variation for things like nepotism). The rest of the student body are there as "filler", to get sufficient headcount for the university to fund its gargantuan bureaucracy.

    That is something that should be made really clear to non-élite high school students: if you don't get a full scholarship for university, the probability that you will benefit economically from tertiary study is near-zero[1].

    As a result, the median stats for graduates as a whole doesn't impart much information (and since it's much easier to get into, and graduate from, a US university - and since the US university-going cohort is the biggest by far in the West... well, that will drag down the global median).

    It's been my 'rule of thumb' for some time that the median Honours graduate in a moderately-technical discipline can safely be assumed to have an IQ 1.5σ above the population median.

    That's using my jurisdiction's version of 'Honours': roughly the top 10% of the student body (bearing in mind that the student body in technical disciplines is drawn almost exclusively from the top 10% of high school graduates, because there is a cutoff score for entry into technical disciplines).

    Anyhow... although my perception of the median Hons grad might sound like an anecdote, it's based on N reasonably-large: my own Hons class was only 30 people, but I've seen a shitload more than that.

    Part of the sweet, sweet grad school gig as a full-scholarship PhD student, was getting paid 0.375 of a full salary to take 2 or 3 tutorials a week. That was in addition to my priority scholarship, which was full fee relief plus a stipend equal to ~40% of average full time earnings, tax free. I guess in the US system they call it being a TA.

    So I tutored for 4 years; the subjects were first year, third year, and Honours (Hons tutorials were not under the auspices of the university). I also took lectures in a 3rd year subject one semester.

    I've tutored hundreds of undergraduate and Honours students, and the cognitive gap between those who went on to Hons and the rest, was obvious. Very few students are generally 'on the border' between Hons and 'P', and those guys flounder and founder in the Hons class.

    [A note is useful here for the discrepancies in the term 'Honours': in the UK, upwards of 20% of the graduating class are awarded First Class Honours, and 95% of the graduating class gets Second Class II or better. In Australia, only the top 10% of the student body is selected for Honours, and less than 10% of Honours students get a First. A 'First' from a Go8 university pretty much gives you a walk-up start at Oxbridge or the US Ivy League: it certainly gives you 4 years of full 'priority' scholarship, and full fee relief for grad study]

    .

    A second important consideration is the international comparability of the subject matter, which in turn will raise the cognitive bar (in harder 'jurisdictions') for the median graduate/discpline/institution cohorts.

    The US got a running start on credentialism in about the 50s (trying to give still-useful undergraduate university degrees to ex-soldiers), and so the undergraduate syllabi at US 'universities' - even the élite ones - tends toward the generalist.

    As examples: in general you don't get to 'be' an economist, lawyer, or doctor with a US BEc, LLB or MBBS: you have to go to graduate school before you actually start learning anything specific in your discipline (arguably, you could become an economist if you did a BA (Econ) with honors at a place like Stanford... but content-wise you would be a full year behind an Australian BEc graduate from a Go8 university[2]).

    And for other technical fields - Mathematics, Physics, Chemistry, Biology, Engineering... US undergraduate training does not even prepare the student to become a schoolteacher in their area of expertise: grad school is absolutely required to become a 'specialist' in those fields.

    With that said: here in Straya, some of the more prestigious universities are now using a version of the US business model - mostly for courses where the median student is prepared to pay over the odds to participate (e.g., Law,. Medicine). So now you don't do an LLB at Melbourne, finish in the Honours stream and go onto a clerkship at a major firm; you do 'pre-Law' and then a JD.

    More fees, you see: the CEO and the vast bureaucratic machinery has to be paid, sucka!

    End note
    [1] In other words, all of the putative excess return to tertiary education, is captured by the very top of the graduate class.

    [2] It's a hell of a long time ago now, but at my alma mater back in the 1990s the Honours (4th year undergraduate) level macroeconomics course was identical to the Stanford PhD Core Macro course (same textbook; even the same assignment content; taught by a recent Stanford grad). Our course was a 1-semester subject; SU Core Macro was a full-year course.

    As Jon Cougar Mellencamp might have said:


    That's when a First was a First
     
    , @dux.ie
    ENIGMA is a first order coarse grain estimation simply working off the bell curve mathematics and I need that to estimate the migrant average group IQ (MAGIQ) rather than plucking numbers from thin air. The implicit assumptions are that there are complete meritocracy and frictionless university student selection. If the deviations from the assumptions are known then ENIGMA could be modified to accomodate that. For example India seems to have formal quota for the minorities but I do not have those data.

    Immigrant group data like that from IAB usually have the data on the fraction that have university degrees. Pivoting on that and with the ENIGMA value and doing the bell curve maths again can give a ball park figure for the immigrant group IQ. For example the Becker IQ for India is 83.23 and OECD data for the national fraction of gradute is 5.64% that give ENIGMA of 107.02, and the fraction of Indian immigrants to UK with degree (FracHi) from IAB is 45.58% that gives the Indian migrant group IQ to UK MAGIQ=105.35, a bit higher than the CAT estimation of the IQ of the children of Indian immigrants but much higher than the source country national average IQ. May be I should have factored in the effects of the Indian quota system. The effects of the migrant selection policy are more evident for the Nigerian with IQdb=70.59, from UNICEF UniA=7%, from IAB FracHi=72.28% (!), giving MAGIC=101.6, a jump of 30.1 IQ points. For HKG the IQdb for China is used as proxy and the value could be a bit too high.

    Host UK
    Rank IQcat MAGIQ ENIGMA IQdb UniA FracHi Country
    23 99.8 105.35 107.02 83.23 5.64 45.58 IND
    29 93.4 101.6 92.73 70.59 7.0 72.28(!) NGA Nigeria
    34 93.5 97.27 105.53 79.94 4.4 29.1 PAK
    37 94.7 93.37 104.49 74.56 2.3 22.92 BGD
    42 94.0 85.46 94.28 73.6 8.4 27.82 JAM
    6 107.5 117.21 120.46 106.27 17.21 41.42 HKG

    The Table A1.1 you referred to has zero decimal places and the rounding off error could be significant for low percentages. An 2015 OECD report which I cannot locate now has at least one decimal place data and the values seem to be larger. Some diff

    EAG17|EAG15?|DIF|Country
    31|44.0|-13.0|AUS
    16|20.7|-4.7|AUT
    37|39.1|-2.1|BEL
    22|33.8|-11.8|CZE
    32|49.8|-17.8|DNK

    Furthermore the data on OECD partners cannot be trusted. For example the data for China at 3% (which gives ENIGMA=133.25) has not been updated for quite some time, the EAG15?? listed 26.4% (ENIGMA=115) seems to be more reasonable.

    > 92k is the number who wrote the entrance exam
    That is why I qualified that with "unless it is in a very high IQ region". Actually the 0.251 number was not that high, a more recent UniA seems to be .238 compares to the old 0.162 , that put the new ENIGMA=99.2 even though IQdb added 0.51 IQ point to that from IQLynn. Dispite what the American thinks of the Mexican immigrants, for UK it seems to be a different picture due to the selection policies,

    Rank MAGIQ ENIGMA IQdb UniA FracHi Country
    9 113.74 99.2 88.51 23.8 83.38(!!) MEX
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  22. jim jones says:
    @dearieme
    Interesting idea. I propose a careful trial - begin with the LSE. There would be so little to lose.

    Black MPs in the UK are always complaining about the lack of ethnic students accepted by Oxbridge:

    https://www.theguardian.com/education/2017/oct/19/oxford-accused-of-social-apartheid-as-colleges-admit-no-black-students

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  23. res says:
    @dux.ie
    Part of the attribute of people with university degree is signalling that they are a cut above the average. If more people get university degrees then that might not worth the papers that certified them. In USA the NewYorkFed data showed that for some studies the percentages of university graduates working in jobs that do not require degrees (under-employment) are very high, for example CriminalJustice 75.5%, PerformingArts 65.5%, LiberalArts 57%, MassMedia 56.1%, Sociology 52.1%, EthnicStudies 50.5%, etc. And they might already have acquired hefty student loans to achieve that.

    Another is the international competitiveness of the country's graduates. The nominal IQ of university graduate is 115, but many countries trying to cut corners and devalueing their worth. The international competitiveness of the graduate is dependent on the average national IQ and the fraction of the population with degrees which I measured with ENIGMA (Estimated Nominal IQ of Graduate from Multiple Attributes). If the average national IQ is low then the country should reduce the fraction of the population with degrees, otherwise their degrees are less worthy internationally. Example,

    Rank ENIGMA IQLynn UniA Frac115 SRatio Country
    2 118.04 102 0.142 0.193 1.36 ITA
    3 117.60 100 0.120 0.159 1.32 AUT
    4 114.84 106 0.278 0.274 0.99 KOR
    5 114.51 105 0.263 0.252 0.96 JPN
    6 114.49 99 0.151 0.143 0.95 DEU
    24 108.47 100 0.286 0.159 0.55 GBR
    25 108.29 99 0.268 0.143 0.53 CAN
    33 105.54 98 0.308 0.129 0.42 USA
    38 102.78 88 0.162 0.036 0.22 MEX

    The national average IQ data from Lynn for Mexico is 88. From OECD data the fraction of Mexican with degrees is 0.162, that gives ENIGMA=102.8 and signalling ratio SRatio=0.22, already way below the nominal ENIGMA value of 115.

    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162. Dispite what that person interviewed said, the prestige of that university might already below average withn the country, unless it is in a very high IQ region. To reach ENIGMA=115 that fraction should be 0.036

    I like your ENIGMA methodology. Do you expand on it anywhere (e.g. the full table of calculations)? The only thing I was able to find was this very interesting comment (and followups) at West Hunter: https://westhunt.wordpress.com/2017/06/17/african-variation/#comment-92877

    Have you tried cross checking your Frac115 numbers with PISA results?

    Could you elaborate a bit more on which OECD data (e.g. year, age range, variables included in UniA) you used and which source had the data with three significant digits? The best I am finding is Table A1.1 and A1.2 of Education at a Glance: http://www.oecd.org/education/education-at-a-glance-19991487.htm
    Annex 3 has more detail including a link to the spreadsheet below: http://www.oecd.org/edu/skills-beyond-school/EAG2017-Annex-3.pdf
    Here are the 2016 Tables in Excel: http://dx.doi.org/10.1787/888933562885

    I assume ENIGMA is just the UniA converted to an IQ delta (Z-score * 15) and added to IQLynn?

    The big issue I see is that I think you are using the Table A1.1 data (ages 25-64) while Table A1.2 (the spreadsheet version is much better than the PDF because it breaks out the different levels of tertiary education) shows that for a country like Mexico UniA varies by about a factor of two between the 25-34 and 55-64 age groups. That effect is probably partially canceled by the Flynn Effect, but I am not sure how to correct for those.

    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162.

    I think you are comparing apples and oranges here. 92k is the number who wrote the entrance exam. Your other statistic is based on total population. What is the percentage of the student age population who applied to the university? The data I see in Table A1.1 for Mexico (exclusive categories) is 19% complete secondary and 15% have a bachelors or above. So even if everyone who completes secondary (~34%) applies you need to divide your 0.251 number by 3.

    P.S. The data tables in the PDF document do not include a description of the data annotations, so here they are from the spreadsheet.

    [MORE]

    Symbols for missing data

    m Data are not available
    a Data are not applicable because the category does not apply
    x Data included in another category
    d Includes data from another category
    0 Magnitude is either negligible or zero
    c There are too few observations to provide reliable estimates.
    r Values are below a certain reliability threshold and should be interpreted with caution
    b Break in the time series

    Read More
    • Replies: @dux.ie
    ENIGMA is just a fancy name for simple bell curve mathematics so far only a first order coarse grain estimation based on average national IQ and the fraction of population with university degrees. Refinement is possible but age segmented IQ data are generally not available. No time to elaborate now. Frac115 is just given the average national IQ the fraction of the population with IQ ≥ 115.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  24. gate666 says:
    @Edward
    To which estimates are you referring? These data are much more recent than most of Lynn's estimates, and have a substantially larger sample size (10 times larger, or more, than most of Lynn's studies).

    lynn claimed iq of british indians was around 83.

    Read More
    • Replies: @Edward
    Where? The average IQ in India is somewhere between 82 (Lynn) and 87 (Rushton and Jensen). The only comment of Lynn's that I can find on British Indian IQ is in 'Race Differences in Intelligence: A Global Perspective', in which he states:

    Mean IQs [in India] lie in the range of 81 to 94, with an overall mean of approximately 86. But ethic Indians in Britain obtain a mean of 96 which is within the range of other Caucasoid populations. Their verbal IQ of 89 is depressed, but this is probably because their families are recent immigrants and have not yet mastered the language. The British results suggest that when Indians are reared in an economically developed environment their intelligence level is about the same as that of European Caucasoids.
     
    Of course, this was published in the 1990s, and as I said earlier most of the old studies on the IQs of immigrants to the UK had very small sample sizes. These recent data have by far the largest sample sizes, so an average of ~100 is probably closer to the true value. If we assume that the average IQ in India is around 84, and Indians in Britain perhaps gain around 8-10 points due to better environmental conditions, with the rest of the improvement being due to more intelligent Indians on average emigrating to the UK, then 100 seems perfectly plausible. Neither 83 nor 96 are particularly plausible for British Indians, given that they have the highest average incomes in the UK and are much more likely than any other ethnic group to be employed in professional and managerial occupations. Educationally, they are only outperformed by the Chinese at GCSE and A-Level.

    (The higher average intelligence of Indian immigrants relative to people living in India is, incidentally, likely due to self-selection, not government selection, as citizens from all of our former colonies were able to come here from the end of WWII to the early 1960s, and the vast majority of the Indian students in the data that Dr. Thompson cites are therefore third-generation immigrants.)

    This also works for the other ethnic groups mentioned by Dr. Thompson. If we take a compromise between Lynn's IQ estimates for sub-Saharan Africa (70) and Wicherts' estimates (80), we get an average of 75. Another paper by Wicherts suggests that 75 is actually the most reasonable estimate, incidentally. The average IQ of Black Africans from the CAT data is 93.4. Again, if we assume that immigrants gain 8-10 points due to better environmental conditions, with the rest of the improvement being down to selection, the CAT data seem to be on-the-mark.

    Pakistanis and Bangladeshis (to a lesser extent) have alarmingly high rates of cousin marriage which have long been noted to cause disability, which might explain why their average IQs are significantly lower than that of British Indians. It could also have something to do with the Hindu caste system.

    Note: environmental gains of 8-10 points are in line with the gains that Steve Sailer and others have estimated. If the mean IQ of African-Americans is 85, and of sub-Saharan Africans is 75, then 8-10 points seems about right, given that admixture will bump African-American IQ up by a couple of points and negative selection for intelligence during slavery would have maybe bumped it down by one or two points (as Chuck has estimated).

    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  25. Factorize says:

    300k IQ GWAS.

    Is this a new one? Sometimes it is not always clear if it is hot off the press or not.

    https://www.nature.com/articles/s41467-018-04362-x

    Read More
    • Replies: @res
    I'm not sure if it is new or not either. Anyone?

    There is a great deal of Supplementary Material for that. Supplemental Data 3 has a nice spreadsheet with ~20,000 SNPs and results for this GWAS and three earlier studies.

    Look-ups of independent significant SNPs and tagged SNPs for general cognitive function with intelligence (Hill et al., 2017), educational attainment (Okbay et al., 2016 GWAS) and cognitive function (Sniekers et al., 2017).
     
    , @hyperbola
    Another study that confirms that "cognitive function" is a VERY complex trait at the genetic level - so much so that claims of "prediction" of cognitive function (IQ) by genetic makeup is a fraud.

    Gene-based analyses find 709 genes associated with general cognitive function. Expression levels across the cortex are associated with general cognitive function. Using polygenic scores, up to 4.3% of variance in general cognitive function is predicted in independent samples.
     
    Given that is so, one wonders when diehards like J. Thompson will finally give up on silly hypotheses about genes/race/IQ correlations that have no validity at the individual level.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  26. Factorize says:

    Arrggh!

    This article is highly similar to:

    https://www.biorxiv.org/content/biorxiv/early/2017/08/17/176511.full.pdf

    Might be very helpful if they were to include a reference to the original biorxiv article so that it is clear what is new and what isn’t.

    Read More
    • Replies: @res
    Also see the updated version of that at https://www.biorxiv.org/content/early/2017/08/18/176511
    which was discussed here in http://www.unz.com/jthompson/the-99-steps-of-intelligence-hunters/

    50% more claimed SNPs is an important difference though.

    I can see why Nature might not want to link to biorxiv, but it would be nice for the biorxiv page to add a link to the published paper once it is available.

    Another possibility would be somehow associating the different DOIs. Here 10.1101/176511 and 10.1038/s41467-018-04362-x
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  27. res says:
    @Factorize
    300k IQ GWAS.

    Is this a new one? Sometimes it is not always clear if it is hot off the press or not.
    https://www.nature.com/articles/s41467-018-04362-x

    I’m not sure if it is new or not either. Anyone?

    There is a great deal of Supplementary Material for that. Supplemental Data 3 has a nice spreadsheet with ~20,000 SNPs and results for this GWAS and three earlier studies.

    Look-ups of independent significant SNPs and tagged SNPs for general cognitive function with intelligence (Hill et al., 2017), educational attainment (Okbay et al., 2016 GWAS) and cognitive function (Sniekers et al., 2017).

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  28. res says:
    @Factorize
    Arrggh!

    This article is highly similar to:
    https://www.biorxiv.org/content/biorxiv/early/2017/08/17/176511.full.pdf

    Might be very helpful if they were to include a reference to the original biorxiv article so that it is clear what is new and what isn't.

    Also see the updated version of that at https://www.biorxiv.org/content/early/2017/08/18/176511
    which was discussed here in http://www.unz.com/jthompson/the-99-steps-of-intelligence-hunters/

    50% more claimed SNPs is an important difference though.

    I can see why Nature might not want to link to biorxiv, but it would be nice for the biorxiv page to add a link to the published paper once it is available.

    Another possibility would be somehow associating the different DOIs. Here 10.1101/176511 and 10.1038/s41467-018-04362-x

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  29. @dux.ie
    Part of the attribute of people with university degree is signalling that they are a cut above the average. If more people get university degrees then that might not worth the papers that certified them. In USA the NewYorkFed data showed that for some studies the percentages of university graduates working in jobs that do not require degrees (under-employment) are very high, for example CriminalJustice 75.5%, PerformingArts 65.5%, LiberalArts 57%, MassMedia 56.1%, Sociology 52.1%, EthnicStudies 50.5%, etc. And they might already have acquired hefty student loans to achieve that.

    Another is the international competitiveness of the country's graduates. The nominal IQ of university graduate is 115, but many countries trying to cut corners and devalueing their worth. The international competitiveness of the graduate is dependent on the average national IQ and the fraction of the population with degrees which I measured with ENIGMA (Estimated Nominal IQ of Graduate from Multiple Attributes). If the average national IQ is low then the country should reduce the fraction of the population with degrees, otherwise their degrees are less worthy internationally. Example,

    Rank ENIGMA IQLynn UniA Frac115 SRatio Country
    2 118.04 102 0.142 0.193 1.36 ITA
    3 117.60 100 0.120 0.159 1.32 AUT
    4 114.84 106 0.278 0.274 0.99 KOR
    5 114.51 105 0.263 0.252 0.96 JPN
    6 114.49 99 0.151 0.143 0.95 DEU
    24 108.47 100 0.286 0.159 0.55 GBR
    25 108.29 99 0.268 0.143 0.53 CAN
    33 105.54 98 0.308 0.129 0.42 USA
    38 102.78 88 0.162 0.036 0.22 MEX

    The national average IQ data from Lynn for Mexico is 88. From OECD data the fraction of Mexican with degrees is 0.162, that gives ENIGMA=102.8 and signalling ratio SRatio=0.22, already way below the nominal ENIGMA value of 115.

    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162. Dispite what that person interviewed said, the prestige of that university might already below average withn the country, unless it is in a very high IQ region. To reach ENIGMA=115 that fraction should be 0.036

    The nominal IQ of university graduate is 115

    That aggregate statistic is pretty pointless, since it conceals significant variation by institution and discipline: the median graduate in Physics from Cal Tech or MIT is a different species from the median graduate in Gender Studies from Pomona (and Pomona’s not a bad school).

    Go further down the school pile, and push the discipline further towards “hobby” status, and watch the median student’s IQ drop toward the mean.

    In other words, there are significantly-different sub-populations within the ‘university graduate’ cohort: for some of those sub-populations, 115 is a significant under-estimate.

    And let’s stipulate: the reason why a shitload of graduates work in areas that do not require a degree, is because anyone below the top decile of graduates is not useful in their discipline.

    Nobody outside of the top 5% of the graduating class is getting a discipline-specific job straight out of their undergraduate degree (with teensy tiny variation for things like nepotism). The rest of the student body are there as “filler”, to get sufficient headcount for the university to fund its gargantuan bureaucracy.

    That is something that should be made really clear to non-élite high school students: if you don’t get a full scholarship for university, the probability that you will benefit economically from tertiary study is near-zero[1].

    As a result, the median stats for graduates as a whole doesn’t impart much information (and since it’s much easier to get into, and graduate from, a US university – and since the US university-going cohort is the biggest by far in the West… well, that will drag down the global median).

    It’s been my ‘rule of thumb’ for some time that the median Honours graduate in a moderately-technical discipline can safely be assumed to have an IQ 1.5σ above the population median.

    That’s using my jurisdiction’s version of ‘Honours’: roughly the top 10% of the student body (bearing in mind that the student body in technical disciplines is drawn almost exclusively from the top 10% of high school graduates, because there is a cutoff score for entry into technical disciplines).

    Anyhow… although my perception of the median Hons grad might sound like an anecdote, it’s based on N reasonably-large: my own Hons class was only 30 people, but I’ve seen a shitload more than that.

    Part of the sweet, sweet grad school gig as a full-scholarship PhD student, was getting paid 0.375 of a full salary to take 2 or 3 tutorials a week. That was in addition to my priority scholarship, which was full fee relief plus a stipend equal to ~40% of average full time earnings, tax free. I guess in the US system they call it being a TA.

    So I tutored for 4 years; the subjects were first year, third year, and Honours (Hons tutorials were not under the auspices of the university). I also took lectures in a 3rd year subject one semester.

    I’ve tutored hundreds of undergraduate and Honours students, and the cognitive gap between those who went on to Hons and the rest, was obvious. Very few students are generally ‘on the border’ between Hons and ‘P’, and those guys flounder and founder in the Hons class.

    [A note is useful here for the discrepancies in the term 'Honours': in the UK, upwards of 20% of the graduating class are awarded First Class Honours, and 95% of the graduating class gets Second Class II or better. In Australia, only the top 10% of the student body is selected for Honours, and less than 10% of Honours students get a First. A 'First' from a Go8 university pretty much gives you a walk-up start at Oxbridge or the US Ivy League: it certainly gives you 4 years of full 'priority' scholarship, and full fee relief for grad study]

    .

    A second important consideration is the international comparability of the subject matter, which in turn will raise the cognitive bar (in harder ‘jurisdictions’) for the median graduate/discpline/institution cohorts.

    The US got a running start on credentialism in about the 50s (trying to give still-useful undergraduate university degrees to ex-soldiers), and so the undergraduate syllabi at US ‘universities’ – even the élite ones – tends toward the generalist.

    As examples: in general you don’t get to ‘be’ an economist, lawyer, or doctor with a US BEc, LLB or MBBS: you have to go to graduate school before you actually start learning anything specific in your discipline (arguably, you could become an economist if you did a BA (Econ) with honors at a place like Stanford… but content-wise you would be a full year behind an Australian BEc graduate from a Go8 university[2]).

    And for other technical fields – Mathematics, Physics, Chemistry, Biology, Engineering… US undergraduate training does not even prepare the student to become a schoolteacher in their area of expertise: grad school is absolutely required to become a ‘specialist’ in those fields.

    With that said: here in Straya, some of the more prestigious universities are now using a version of the US business model – mostly for courses where the median student is prepared to pay over the odds to participate (e.g., Law,. Medicine). So now you don’t do an LLB at Melbourne, finish in the Honours stream and go onto a clerkship at a major firm; you do ‘pre-Law’ and then a JD.

    More fees, you see: the CEO and the vast bureaucratic machinery has to be paid, sucka!

    End note
    [1] In other words, all of the putative excess return to tertiary education, is captured by the very top of the graduate class.

    [2] It’s a hell of a long time ago now, but at my alma mater back in the 1990s the Honours (4th year undergraduate) level macroeconomics course was identical to the Stanford PhD Core Macro course (same textbook; even the same assignment content; taught by a recent Stanford grad). Our course was a 1-semester subject; SU Core Macro was a full-year course.

    As Jon Cougar Mellencamp might have said:

    That’s when a First was a First

    Read More
    • Replies: @dux.ie
    >> The nominal IQ of university graduate is 115
    > That aggregate statistic is pretty pointless, since it conceals significant variation by institution and discipline.

    Yes. However the analysis is supposed to apply to groups on average not individual. This is further expanded into estimating the quality of migrant groups from various countries and analytically dispel the common myths that just because the migrants were from countries with low average national IQ that they cannot have on average superb performance in the new host countries. The misinformed myths are further disproved from the US census household incomes data for the migrant groups like the Subcontinent Indians, Nigerian, etc.

    From Wai's data on university entrance SAT score the minimum seems to be SAT score of 650, i.e. IQlike value of 75. The nominal IQ of university graduate of 115 has been generally talk about. Recently I found out that number 115 came out naturally from empirical data as the cutoff point where meritocracy seems to be dominant, separate from the behaviour of the other graduate cluster where many of them believed that IQ does not affect future income and that equality of outcome is important and that ideology actually shape the reality for their cluster and they cannot perceive what is happening in the other meritocratic cluster.

    There are other analysis for the entrance IQ for various majors, http://www.statisticbrain.com/iq-estimates-by-intended-college-major/ The New York Federal Reserve has data on under-employment, i.e. the percentage of university graduates in various majors working in jobs that do not require university degrees, https://www.newyorkfed.org/research/college-labor-market/college-labor-market_compare-majors.html The analysis was based on last year's data.

    Mashing the two datasets showed the distinct seperation of two clusters with different characteristics. Those with entry IQ for the majors less than 115, on average the higher the IQ the more they tended to congregate into majors with higher under-employment rate. A real life behaviour that is very hard to comprehend. Above the cut-off of 115, the higher the entry IQ the lesser the under-employment rate.

    http://tinypic.com/view.php?pic=10n5ph5&s=9

    The next chart shows the reported incomes as tabled by the NewYorkFed. In the major entry IQ less than 115 cluster, statistically the income level is flat with respect to entry IQ, the ideal as avocated by the dominant voice in that cluster. The flocking to the higher under-employment sectors has on average no incentive of getting higher income. The outliers are jobs that deal with life matters and confidentiality. On the other hand in the other meritocratic cluster, statistically the higher the entry IQ the higher the income level. The horizontal bar is the average income for jobs that do not require university degrees.

    http://tinypic.com/view.php?pic=1ik9bm&s=9

    Thus there are some strange real life behaviours associating with the nominal IQ of 115.
    , @James Thompson
    The smart fraction.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  30. dux.ie says:
    @Kratoklastes

    The nominal IQ of university graduate is 115
     
    That aggregate statistic is pretty pointless, since it conceals significant variation by institution and discipline: the median graduate in Physics from Cal Tech or MIT is a different species from the median graduate in Gender Studies from Pomona (and Pomona's not a bad school).

    Go further down the school pile, and push the discipline further towards "hobby" status, and watch the median student's IQ drop toward the mean.

    In other words, there are significantly-different sub-populations within the 'university graduate' cohort: for some of those sub-populations, 115 is a significant under-estimate.

    And let's stipulate: the reason why a shitload of graduates work in areas that do not require a degree, is because anyone below the top decile of graduates is not useful in their discipline.

    Nobody outside of the top 5% of the graduating class is getting a discipline-specific job straight out of their undergraduate degree (with teensy tiny variation for things like nepotism). The rest of the student body are there as "filler", to get sufficient headcount for the university to fund its gargantuan bureaucracy.

    That is something that should be made really clear to non-élite high school students: if you don't get a full scholarship for university, the probability that you will benefit economically from tertiary study is near-zero[1].

    As a result, the median stats for graduates as a whole doesn't impart much information (and since it's much easier to get into, and graduate from, a US university - and since the US university-going cohort is the biggest by far in the West... well, that will drag down the global median).

    It's been my 'rule of thumb' for some time that the median Honours graduate in a moderately-technical discipline can safely be assumed to have an IQ 1.5σ above the population median.

    That's using my jurisdiction's version of 'Honours': roughly the top 10% of the student body (bearing in mind that the student body in technical disciplines is drawn almost exclusively from the top 10% of high school graduates, because there is a cutoff score for entry into technical disciplines).

    Anyhow... although my perception of the median Hons grad might sound like an anecdote, it's based on N reasonably-large: my own Hons class was only 30 people, but I've seen a shitload more than that.

    Part of the sweet, sweet grad school gig as a full-scholarship PhD student, was getting paid 0.375 of a full salary to take 2 or 3 tutorials a week. That was in addition to my priority scholarship, which was full fee relief plus a stipend equal to ~40% of average full time earnings, tax free. I guess in the US system they call it being a TA.

    So I tutored for 4 years; the subjects were first year, third year, and Honours (Hons tutorials were not under the auspices of the university). I also took lectures in a 3rd year subject one semester.

    I've tutored hundreds of undergraduate and Honours students, and the cognitive gap between those who went on to Hons and the rest, was obvious. Very few students are generally 'on the border' between Hons and 'P', and those guys flounder and founder in the Hons class.

    [A note is useful here for the discrepancies in the term 'Honours': in the UK, upwards of 20% of the graduating class are awarded First Class Honours, and 95% of the graduating class gets Second Class II or better. In Australia, only the top 10% of the student body is selected for Honours, and less than 10% of Honours students get a First. A 'First' from a Go8 university pretty much gives you a walk-up start at Oxbridge or the US Ivy League: it certainly gives you 4 years of full 'priority' scholarship, and full fee relief for grad study]

    .

    A second important consideration is the international comparability of the subject matter, which in turn will raise the cognitive bar (in harder 'jurisdictions') for the median graduate/discpline/institution cohorts.

    The US got a running start on credentialism in about the 50s (trying to give still-useful undergraduate university degrees to ex-soldiers), and so the undergraduate syllabi at US 'universities' - even the élite ones - tends toward the generalist.

    As examples: in general you don't get to 'be' an economist, lawyer, or doctor with a US BEc, LLB or MBBS: you have to go to graduate school before you actually start learning anything specific in your discipline (arguably, you could become an economist if you did a BA (Econ) with honors at a place like Stanford... but content-wise you would be a full year behind an Australian BEc graduate from a Go8 university[2]).

    And for other technical fields - Mathematics, Physics, Chemistry, Biology, Engineering... US undergraduate training does not even prepare the student to become a schoolteacher in their area of expertise: grad school is absolutely required to become a 'specialist' in those fields.

    With that said: here in Straya, some of the more prestigious universities are now using a version of the US business model - mostly for courses where the median student is prepared to pay over the odds to participate (e.g., Law,. Medicine). So now you don't do an LLB at Melbourne, finish in the Honours stream and go onto a clerkship at a major firm; you do 'pre-Law' and then a JD.

    More fees, you see: the CEO and the vast bureaucratic machinery has to be paid, sucka!

    End note
    [1] In other words, all of the putative excess return to tertiary education, is captured by the very top of the graduate class.

    [2] It's a hell of a long time ago now, but at my alma mater back in the 1990s the Honours (4th year undergraduate) level macroeconomics course was identical to the Stanford PhD Core Macro course (same textbook; even the same assignment content; taught by a recent Stanford grad). Our course was a 1-semester subject; SU Core Macro was a full-year course.

    As Jon Cougar Mellencamp might have said:


    That's when a First was a First
     

    >> The nominal IQ of university graduate is 115
    > That aggregate statistic is pretty pointless, since it conceals significant variation by institution and discipline.

    Yes. However the analysis is supposed to apply to groups on average not individual. This is further expanded into estimating the quality of migrant groups from various countries and analytically dispel the common myths that just because the migrants were from countries with low average national IQ that they cannot have on average superb performance in the new host countries. The misinformed myths are further disproved from the US census household incomes data for the migrant groups like the Subcontinent Indians, Nigerian, etc.

    From Wai’s data on university entrance SAT score the minimum seems to be SAT score of 650, i.e. IQlike value of 75. The nominal IQ of university graduate of 115 has been generally talk about. Recently I found out that number 115 came out naturally from empirical data as the cutoff point where meritocracy seems to be dominant, separate from the behaviour of the other graduate cluster where many of them believed that IQ does not affect future income and that equality of outcome is important and that ideology actually shape the reality for their cluster and they cannot perceive what is happening in the other meritocratic cluster.

    There are other analysis for the entrance IQ for various majors, http://www.statisticbrain.com/iq-estimates-by-intended-college-major/ The New York Federal Reserve has data on under-employment, i.e. the percentage of university graduates in various majors working in jobs that do not require university degrees, https://www.newyorkfed.org/research/college-labor-market/college-labor-market_compare-majors.html The analysis was based on last year’s data.

    Mashing the two datasets showed the distinct seperation of two clusters with different characteristics. Those with entry IQ for the majors less than 115, on average the higher the IQ the more they tended to congregate into majors with higher under-employment rate. A real life behaviour that is very hard to comprehend. Above the cut-off of 115, the higher the entry IQ the lesser the under-employment rate.

    http://tinypic.com/view.php?pic=10n5ph5&s=9

    The next chart shows the reported incomes as tabled by the NewYorkFed. In the major entry IQ less than 115 cluster, statistically the income level is flat with respect to entry IQ, the ideal as avocated by the dominant voice in that cluster. The flocking to the higher under-employment sectors has on average no incentive of getting higher income. The outliers are jobs that deal with life matters and confidentiality. On the other hand in the other meritocratic cluster, statistically the higher the entry IQ the higher the income level. The horizontal bar is the average income for jobs that do not require university degrees.

    http://tinypic.com/view.php?pic=1ik9bm&s=9

    Thus there are some strange real life behaviours associating with the nominal IQ of 115.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  31. dux.ie says:
    @res
    I like your ENIGMA methodology. Do you expand on it anywhere (e.g. the full table of calculations)? The only thing I was able to find was this very interesting comment (and followups) at West Hunter: https://westhunt.wordpress.com/2017/06/17/african-variation/#comment-92877

    Have you tried cross checking your Frac115 numbers with PISA results?

    Could you elaborate a bit more on which OECD data (e.g. year, age range, variables included in UniA) you used and which source had the data with three significant digits? The best I am finding is Table A1.1 and A1.2 of Education at a Glance: http://www.oecd.org/education/education-at-a-glance-19991487.htm
    Annex 3 has more detail including a link to the spreadsheet below: www.oecd.org/edu/skills-beyond-school/EAG2017-Annex-3.pdf
    Here are the 2016 Tables in Excel: http://dx.doi.org/10.1787/888933562885

    I assume ENIGMA is just the UniA converted to an IQ delta (Z-score * 15) and added to IQLynn?

    The big issue I see is that I think you are using the Table A1.1 data (ages 25-64) while Table A1.2 (the spreadsheet version is much better than the PDF because it breaks out the different levels of tertiary education) shows that for a country like Mexico UniA varies by about a factor of two between the 25-34 and 55-64 age groups. That effect is probably partially canceled by the Flynn Effect, but I am not sure how to correct for those.


    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162.
     
    I think you are comparing apples and oranges here. 92k is the number who wrote the entrance exam. Your other statistic is based on total population. What is the percentage of the student age population who applied to the university? The data I see in Table A1.1 for Mexico (exclusive categories) is 19% complete secondary and 15% have a bachelors or above. So even if everyone who completes secondary (~34%) applies you need to divide your 0.251 number by 3.

    P.S. The data tables in the PDF document do not include a description of the data annotations, so here they are from the spreadsheet.

    Symbols for missing data

    m Data are not available
    a Data are not applicable because the category does not apply
    x Data included in another category
    d Includes data from another category
    0 Magnitude is either negligible or zero
    c There are too few observations to provide reliable estimates.
    r Values are below a certain reliability threshold and should be interpreted with caution
    b Break in the time series

    ENIGMA is just a fancy name for simple bell curve mathematics so far only a first order coarse grain estimation based on average national IQ and the fraction of population with university degrees. Refinement is possible but age segmented IQ data are generally not available. No time to elaborate now. Frac115 is just given the average national IQ the fraction of the population with IQ ≥ 115.

    Read More
    • Replies: @Jim Bob Lassiter
    Look, I'll be the first to admit that I have a very disorganized academic past culminating in a US BSBA from a second tier state supported university (that still had a borderline first tier B-School in the mid-seventies) that I never really needed in the absolute technical academic proficiency sense of the word. I did learn to use a slide rule quite proficiently in a calculus course taught by an Economics professor and used same quite proficiently in Banking and Finance classes. Made a D in Statistics taught by a Greek (I don't mean a frat rat either) whose spoken English was so bad, he disappeared after one semester. Repeated course with a B under Chairman of Economics Dept. All of this was occurring as the transition from slide rule to H-P calculators was looming up front and close and computers were still a main frame a hundred miles away and FORTRAN uploads from punched card stacks were transmitted late at night by grad students when AT&T long distance phone rates were (relatively) low.

    So going back to my question (19) about Mexico which generated the response from dux.ie (21) with the chart a few days back, I'm still scratchin' my little low IQ Columbo head about a couple of items after spending a fair amount of time just decompressing the chart's columns and lining them up with their respective headings:

    RE: country names -- AUT = Austria or Australia?? DEU= Germany ??

    RE: number row for USA-- How is it that 30.8% of the population of the USA has some sort of college degree, when only 12.9% of the population is admitted to a college to start with? Surely something has flown over my head here.

    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  32. @Kratoklastes

    The nominal IQ of university graduate is 115
     
    That aggregate statistic is pretty pointless, since it conceals significant variation by institution and discipline: the median graduate in Physics from Cal Tech or MIT is a different species from the median graduate in Gender Studies from Pomona (and Pomona's not a bad school).

    Go further down the school pile, and push the discipline further towards "hobby" status, and watch the median student's IQ drop toward the mean.

    In other words, there are significantly-different sub-populations within the 'university graduate' cohort: for some of those sub-populations, 115 is a significant under-estimate.

    And let's stipulate: the reason why a shitload of graduates work in areas that do not require a degree, is because anyone below the top decile of graduates is not useful in their discipline.

    Nobody outside of the top 5% of the graduating class is getting a discipline-specific job straight out of their undergraduate degree (with teensy tiny variation for things like nepotism). The rest of the student body are there as "filler", to get sufficient headcount for the university to fund its gargantuan bureaucracy.

    That is something that should be made really clear to non-élite high school students: if you don't get a full scholarship for university, the probability that you will benefit economically from tertiary study is near-zero[1].

    As a result, the median stats for graduates as a whole doesn't impart much information (and since it's much easier to get into, and graduate from, a US university - and since the US university-going cohort is the biggest by far in the West... well, that will drag down the global median).

    It's been my 'rule of thumb' for some time that the median Honours graduate in a moderately-technical discipline can safely be assumed to have an IQ 1.5σ above the population median.

    That's using my jurisdiction's version of 'Honours': roughly the top 10% of the student body (bearing in mind that the student body in technical disciplines is drawn almost exclusively from the top 10% of high school graduates, because there is a cutoff score for entry into technical disciplines).

    Anyhow... although my perception of the median Hons grad might sound like an anecdote, it's based on N reasonably-large: my own Hons class was only 30 people, but I've seen a shitload more than that.

    Part of the sweet, sweet grad school gig as a full-scholarship PhD student, was getting paid 0.375 of a full salary to take 2 or 3 tutorials a week. That was in addition to my priority scholarship, which was full fee relief plus a stipend equal to ~40% of average full time earnings, tax free. I guess in the US system they call it being a TA.

    So I tutored for 4 years; the subjects were first year, third year, and Honours (Hons tutorials were not under the auspices of the university). I also took lectures in a 3rd year subject one semester.

    I've tutored hundreds of undergraduate and Honours students, and the cognitive gap between those who went on to Hons and the rest, was obvious. Very few students are generally 'on the border' between Hons and 'P', and those guys flounder and founder in the Hons class.

    [A note is useful here for the discrepancies in the term 'Honours': in the UK, upwards of 20% of the graduating class are awarded First Class Honours, and 95% of the graduating class gets Second Class II or better. In Australia, only the top 10% of the student body is selected for Honours, and less than 10% of Honours students get a First. A 'First' from a Go8 university pretty much gives you a walk-up start at Oxbridge or the US Ivy League: it certainly gives you 4 years of full 'priority' scholarship, and full fee relief for grad study]

    .

    A second important consideration is the international comparability of the subject matter, which in turn will raise the cognitive bar (in harder 'jurisdictions') for the median graduate/discpline/institution cohorts.

    The US got a running start on credentialism in about the 50s (trying to give still-useful undergraduate university degrees to ex-soldiers), and so the undergraduate syllabi at US 'universities' - even the élite ones - tends toward the generalist.

    As examples: in general you don't get to 'be' an economist, lawyer, or doctor with a US BEc, LLB or MBBS: you have to go to graduate school before you actually start learning anything specific in your discipline (arguably, you could become an economist if you did a BA (Econ) with honors at a place like Stanford... but content-wise you would be a full year behind an Australian BEc graduate from a Go8 university[2]).

    And for other technical fields - Mathematics, Physics, Chemistry, Biology, Engineering... US undergraduate training does not even prepare the student to become a schoolteacher in their area of expertise: grad school is absolutely required to become a 'specialist' in those fields.

    With that said: here in Straya, some of the more prestigious universities are now using a version of the US business model - mostly for courses where the median student is prepared to pay over the odds to participate (e.g., Law,. Medicine). So now you don't do an LLB at Melbourne, finish in the Honours stream and go onto a clerkship at a major firm; you do 'pre-Law' and then a JD.

    More fees, you see: the CEO and the vast bureaucratic machinery has to be paid, sucka!

    End note
    [1] In other words, all of the putative excess return to tertiary education, is captured by the very top of the graduate class.

    [2] It's a hell of a long time ago now, but at my alma mater back in the 1990s the Honours (4th year undergraduate) level macroeconomics course was identical to the Stanford PhD Core Macro course (same textbook; even the same assignment content; taught by a recent Stanford grad). Our course was a 1-semester subject; SU Core Macro was a full-year course.

    As Jon Cougar Mellencamp might have said:


    That's when a First was a First
     

    The smart fraction.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  33. Where do these CAT test figures come from, and what percentage of UK pupils take the CAT test? Don’t recall any of my brood mentioning it.

    Re Oxbridge, I know several 3-A kids who didn’t apply.

    In some subjects there must be a deal of self-selection – I believe something like a third of applicants for medical school are Indian or Pakistani by ethnicity.

    The tendency of some universities to filter out pre-A-level applicants (where demand is high) by GCSE grades will tend to disadvantage boys. Cardiff, for instance, with 10 applicants for every med school place, uses a scoring system for selecting interview candidates based on a child’s top nine GCSE results. My impression is that boys are less likely than girls to get top grades in subjects which don’t interest them.

    “When applying to Medicine, your top nine GCSEs will be scored using a points system. These nine must include English Language (and/or Welsh First Language), the Sciences and Mathematics. We award 3 points for A* or grade 9, 2 points for an A or grade 8/7 and 1 point for a B or grade 6…Applications will be ordered according to their score and a cut-off point is decided. The cut-off score can change from year to year as it depends on the overall standard of applications we get in any one year. Over the last five years the cut-off score has ranged between 22 to 26 points”

    You can see that an awful lot of kids with four, five, six A* grades won’t make the cut even if those grades include Biology, Chemistry, Maths, Physics, English/Welsh. In the 26-point year 8 A* and one A were needed simply to get an interview!

    Read More
    • Replies: @Edward
    Medical school applications are indeed a brutal process, but given the number of applicants, this is entirely understandable. For Medicine at Oxford, to even get an interview the vast majority of your GCSE grades will have to be A*s. Incidentally, medical schools also use the UKCAT (which is essentially a full-scale IQ test, measuring verbal, abstract, quantitative and logical reasoning ability) or the BMAT (part of which involves testing for logical reasoning ability) as part of the process.

    As for the CAT, I remember taking it myself, and the whole of the year group took it. If I recall correctly, we had to agree to allow Durham University to use our data. However, we have even better information on the CAT courtesy of Ian Deary and colleagues at the University of Edinburgh, who used it in their landmark 2007 study which looked at the correlation between general intelligence and GCSE grades (it was astoundingly high, at 0.81). They describe the CAT in the study:


    The CAT is the most widely used test of reasoning abilities in the UK, with close to one million school students assessed each academic year... CAT2E has 10 separate subtests, which are aggregated into three batteries, providing standardised measures of verbal, quantitative, and nonverbal reasoning abilities. An average of the three standardised scores, the mean CAT score, is also calculated...The results are most often reported as standard age scores, with a mean of 100 and standard deviation of 15. The tests have very high levels of reliability.
     
    https://www.sciencedirect.com/science/article/pii/S0160289606000171
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  34. Edward says:
    @gate666
    lynn claimed iq of british indians was around 83.

    Where? The average IQ in India is somewhere between 82 (Lynn) and 87 (Rushton and Jensen). The only comment of Lynn’s that I can find on British Indian IQ is in ‘Race Differences in Intelligence: A Global Perspective’, in which he states:

    Mean IQs [in India] lie in the range of 81 to 94, with an overall mean of approximately 86. But ethic Indians in Britain obtain a mean of 96 which is within the range of other Caucasoid populations. Their verbal IQ of 89 is depressed, but this is probably because their families are recent immigrants and have not yet mastered the language. The British results suggest that when Indians are reared in an economically developed environment their intelligence level is about the same as that of European Caucasoids.

    Of course, this was published in the 1990s, and as I said earlier most of the old studies on the IQs of immigrants to the UK had very small sample sizes. These recent data have by far the largest sample sizes, so an average of ~100 is probably closer to the true value. If we assume that the average IQ in India is around 84, and Indians in Britain perhaps gain around 8-10 points due to better environmental conditions, with the rest of the improvement being due to more intelligent Indians on average emigrating to the UK, then 100 seems perfectly plausible. Neither 83 nor 96 are particularly plausible for British Indians, given that they have the highest average incomes in the UK and are much more likely than any other ethnic group to be employed in professional and managerial occupations. Educationally, they are only outperformed by the Chinese at GCSE and A-Level.

    (The higher average intelligence of Indian immigrants relative to people living in India is, incidentally, likely due to self-selection, not government selection, as citizens from all of our former colonies were able to come here from the end of WWII to the early 1960s, and the vast majority of the Indian students in the data that Dr. Thompson cites are therefore third-generation immigrants.)

    This also works for the other ethnic groups mentioned by Dr. Thompson. If we take a compromise between Lynn’s IQ estimates for sub-Saharan Africa (70) and Wicherts’ estimates (80), we get an average of 75. Another paper by Wicherts suggests that 75 is actually the most reasonable estimate, incidentally. The average IQ of Black Africans from the CAT data is 93.4. Again, if we assume that immigrants gain 8-10 points due to better environmental conditions, with the rest of the improvement being down to selection, the CAT data seem to be on-the-mark.

    Pakistanis and Bangladeshis (to a lesser extent) have alarmingly high rates of cousin marriage which have long been noted to cause disability, which might explain why their average IQs are significantly lower than that of British Indians. It could also have something to do with the Hindu caste system.

    Note: environmental gains of 8-10 points are in line with the gains that Steve Sailer and others have estimated. If the mean IQ of African-Americans is 85, and of sub-Saharan Africans is 75, then 8-10 points seems about right, given that admixture will bump African-American IQ up by a couple of points and negative selection for intelligence during slavery would have maybe bumped it down by one or two points (as Chuck has estimated).

    Read More
    • Replies: @James Thompson
    I think that more recent data should be brought into the picture.
    http://www.unz.com/jthompson/the-heterogeneous-states-of-india

    Selecting who comes to England may be more important than what England contributes to those who arrive.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  35. Edward says:
    @YetAnotherAnon
    Where do these CAT test figures come from, and what percentage of UK pupils take the CAT test? Don't recall any of my brood mentioning it.

    Re Oxbridge, I know several 3-A kids who didn't apply.

    In some subjects there must be a deal of self-selection - I believe something like a third of applicants for medical school are Indian or Pakistani by ethnicity.

    The tendency of some universities to filter out pre-A-level applicants (where demand is high) by GCSE grades will tend to disadvantage boys. Cardiff, for instance, with 10 applicants for every med school place, uses a scoring system for selecting interview candidates based on a child's top nine GCSE results. My impression is that boys are less likely than girls to get top grades in subjects which don't interest them.


    "When applying to Medicine, your top nine GCSEs will be scored using a points system. These nine must include English Language (and/or Welsh First Language), the Sciences and Mathematics. We award 3 points for A* or grade 9, 2 points for an A or grade 8/7 and 1 point for a B or grade 6...Applications will be ordered according to their score and a cut-off point is decided. The cut-off score can change from year to year as it depends on the overall standard of applications we get in any one year. Over the last five years the cut-off score has ranged between 22 to 26 points"
     
    You can see that an awful lot of kids with four, five, six A* grades won't make the cut even if those grades include Biology, Chemistry, Maths, Physics, English/Welsh. In the 26-point year 8 A* and one A were needed simply to get an interview!

    Medical school applications are indeed a brutal process, but given the number of applicants, this is entirely understandable. For Medicine at Oxford, to even get an interview the vast majority of your GCSE grades will have to be A*s. Incidentally, medical schools also use the UKCAT (which is essentially a full-scale IQ test, measuring verbal, abstract, quantitative and logical reasoning ability) or the BMAT (part of which involves testing for logical reasoning ability) as part of the process.

    As for the CAT, I remember taking it myself, and the whole of the year group took it. If I recall correctly, we had to agree to allow Durham University to use our data. However, we have even better information on the CAT courtesy of Ian Deary and colleagues at the University of Edinburgh, who used it in their landmark 2007 study which looked at the correlation between general intelligence and GCSE grades (it was astoundingly high, at 0.81). They describe the CAT in the study:

    The CAT is the most widely used test of reasoning abilities in the UK, with close to one million school students assessed each academic year… CAT2E has 10 separate subtests, which are aggregated into three batteries, providing standardised measures of verbal, quantitative, and nonverbal reasoning abilities. An average of the three standardised scores, the mean CAT score, is also calculated…The results are most often reported as standard age scores, with a mean of 100 and standard deviation of 15. The tests have very high levels of reliability.

    https://www.sciencedirect.com/science/article/pii/S0160289606000171

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  36. @Edward
    Where? The average IQ in India is somewhere between 82 (Lynn) and 87 (Rushton and Jensen). The only comment of Lynn's that I can find on British Indian IQ is in 'Race Differences in Intelligence: A Global Perspective', in which he states:

    Mean IQs [in India] lie in the range of 81 to 94, with an overall mean of approximately 86. But ethic Indians in Britain obtain a mean of 96 which is within the range of other Caucasoid populations. Their verbal IQ of 89 is depressed, but this is probably because their families are recent immigrants and have not yet mastered the language. The British results suggest that when Indians are reared in an economically developed environment their intelligence level is about the same as that of European Caucasoids.
     
    Of course, this was published in the 1990s, and as I said earlier most of the old studies on the IQs of immigrants to the UK had very small sample sizes. These recent data have by far the largest sample sizes, so an average of ~100 is probably closer to the true value. If we assume that the average IQ in India is around 84, and Indians in Britain perhaps gain around 8-10 points due to better environmental conditions, with the rest of the improvement being due to more intelligent Indians on average emigrating to the UK, then 100 seems perfectly plausible. Neither 83 nor 96 are particularly plausible for British Indians, given that they have the highest average incomes in the UK and are much more likely than any other ethnic group to be employed in professional and managerial occupations. Educationally, they are only outperformed by the Chinese at GCSE and A-Level.

    (The higher average intelligence of Indian immigrants relative to people living in India is, incidentally, likely due to self-selection, not government selection, as citizens from all of our former colonies were able to come here from the end of WWII to the early 1960s, and the vast majority of the Indian students in the data that Dr. Thompson cites are therefore third-generation immigrants.)

    This also works for the other ethnic groups mentioned by Dr. Thompson. If we take a compromise between Lynn's IQ estimates for sub-Saharan Africa (70) and Wicherts' estimates (80), we get an average of 75. Another paper by Wicherts suggests that 75 is actually the most reasonable estimate, incidentally. The average IQ of Black Africans from the CAT data is 93.4. Again, if we assume that immigrants gain 8-10 points due to better environmental conditions, with the rest of the improvement being down to selection, the CAT data seem to be on-the-mark.

    Pakistanis and Bangladeshis (to a lesser extent) have alarmingly high rates of cousin marriage which have long been noted to cause disability, which might explain why their average IQs are significantly lower than that of British Indians. It could also have something to do with the Hindu caste system.

    Note: environmental gains of 8-10 points are in line with the gains that Steve Sailer and others have estimated. If the mean IQ of African-Americans is 85, and of sub-Saharan Africans is 75, then 8-10 points seems about right, given that admixture will bump African-American IQ up by a couple of points and negative selection for intelligence during slavery would have maybe bumped it down by one or two points (as Chuck has estimated).

    I think that more recent data should be brought into the picture.

    http://www.unz.com/jthompson/the-heterogeneous-states-of-india

    Selecting who comes to England may be more important than what England contributes to those who arrive.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  37. dux.ie says:
    @dux.ie
    Part of the attribute of people with university degree is signalling that they are a cut above the average. If more people get university degrees then that might not worth the papers that certified them. In USA the NewYorkFed data showed that for some studies the percentages of university graduates working in jobs that do not require degrees (under-employment) are very high, for example CriminalJustice 75.5%, PerformingArts 65.5%, LiberalArts 57%, MassMedia 56.1%, Sociology 52.1%, EthnicStudies 50.5%, etc. And they might already have acquired hefty student loans to achieve that.

    Another is the international competitiveness of the country's graduates. The nominal IQ of university graduate is 115, but many countries trying to cut corners and devalueing their worth. The international competitiveness of the graduate is dependent on the average national IQ and the fraction of the population with degrees which I measured with ENIGMA (Estimated Nominal IQ of Graduate from Multiple Attributes). If the average national IQ is low then the country should reduce the fraction of the population with degrees, otherwise their degrees are less worthy internationally. Example,

    Rank ENIGMA IQLynn UniA Frac115 SRatio Country
    2 118.04 102 0.142 0.193 1.36 ITA
    3 117.60 100 0.120 0.159 1.32 AUT
    4 114.84 106 0.278 0.274 0.99 KOR
    5 114.51 105 0.263 0.252 0.96 JPN
    6 114.49 99 0.151 0.143 0.95 DEU
    24 108.47 100 0.286 0.159 0.55 GBR
    25 108.29 99 0.268 0.143 0.53 CAN
    33 105.54 98 0.308 0.129 0.42 USA
    38 102.78 88 0.162 0.036 0.22 MEX

    The national average IQ data from Lynn for Mexico is 88. From OECD data the fraction of Mexican with degrees is 0.162, that gives ENIGMA=102.8 and signalling ratio SRatio=0.22, already way below the nominal ENIGMA value of 115.

    For that particular university, 24K/92K=0.251, that is already above the national average of 0.162. Dispite what that person interviewed said, the prestige of that university might already below average withn the country, unless it is in a very high IQ region. To reach ENIGMA=115 that fraction should be 0.036

    ENIGMA is a first order coarse grain estimation simply working off the bell curve mathematics and I need that to estimate the migrant average group IQ (MAGIQ) rather than plucking numbers from thin air. The implicit assumptions are that there are complete meritocracy and frictionless university student selection. If the deviations from the assumptions are known then ENIGMA could be modified to accomodate that. For example India seems to have formal quota for the minorities but I do not have those data.

    Immigrant group data like that from IAB usually have the data on the fraction that have university degrees. Pivoting on that and with the ENIGMA value and doing the bell curve maths again can give a ball park figure for the immigrant group IQ. For example the Becker IQ for India is 83.23 and OECD data for the national fraction of gradute is 5.64% that give ENIGMA of 107.02, and the fraction of Indian immigrants to UK with degree (FracHi) from IAB is 45.58% that gives the Indian migrant group IQ to UK MAGIQ=105.35, a bit higher than the CAT estimation of the IQ of the children of Indian immigrants but much higher than the source country national average IQ. May be I should have factored in the effects of the Indian quota system. The effects of the migrant selection policy are more evident for the Nigerian with IQdb=70.59, from UNICEF UniA=7%, from IAB FracHi=72.28% (!), giving MAGIC=101.6, a jump of 30.1 IQ points. For HKG the IQdb for China is used as proxy and the value could be a bit too high.

    Host UK
    Rank IQcat MAGIQ ENIGMA IQdb UniA FracHi Country
    23 99.8 105.35 107.02 83.23 5.64 45.58 IND
    29 93.4 101.6 92.73 70.59 7.0 72.28(!) NGA Nigeria
    34 93.5 97.27 105.53 79.94 4.4 29.1 PAK
    37 94.7 93.37 104.49 74.56 2.3 22.92 BGD
    42 94.0 85.46 94.28 73.6 8.4 27.82 JAM
    6 107.5 117.21 120.46 106.27 17.21 41.42 HKG

    The Table A1.1 you referred to has zero decimal places and the rounding off error could be significant for low percentages. An 2015 OECD report which I cannot locate now has at least one decimal place data and the values seem to be larger. Some diff

    EAG17|EAG15?|DIF|Country
    31|44.0|-13.0|AUS
    16|20.7|-4.7|AUT
    37|39.1|-2.1|BEL
    22|33.8|-11.8|CZE
    32|49.8|-17.8|DNK

    Furthermore the data on OECD partners cannot be trusted. For example the data for China at 3% (which gives ENIGMA=133.25) has not been updated for quite some time, the EAG15?? listed 26.4% (ENIGMA=115) seems to be more reasonable.

    > 92k is the number who wrote the entrance exam
    That is why I qualified that with “unless it is in a very high IQ region”. Actually the 0.251 number was not that high, a more recent UniA seems to be .238 compares to the old 0.162 , that put the new ENIGMA=99.2 even though IQdb added 0.51 IQ point to that from IQLynn. Dispite what the American thinks of the Mexican immigrants, for UK it seems to be a different picture due to the selection policies,

    Rank MAGIQ ENIGMA IQdb UniA FracHi Country
    9 113.74 99.2 88.51 23.8 83.38(!!) MEX

    Read More
    • Replies: @res

    Immigrant group data like that from IAB usually have the data on the fraction that have university degrees.
     
    My understanding is that the IAB Brain Drain data uses "some tertiary education" rather than university degree. From their methodological note at http://www.iab.de/en/daten/iab-brain-drain-data.aspx

    Educational categories. We distinguish three levels of education: primary (low skilled: includes lower secondary, primary and no schooling); secondary (mediumskilled: high-school leaving certificate or equivalent) and tertiary education (highskilled: higher than high-school leaving certificate or equivalent).
     
    If I understand correctly that is equivalent to "some college." Do you know differently? Table A1.1 has the categories "Post-secondary non-tertiary" and "Short-cycle tertiary." Here is some documentation for what those terms mean: https://www.oecd-ilibrary.org/education/isced-2011-operational-manual/isced-2011-level-5-short-cycle-tertiary-education_9789264228368-10-en https://stats.oecd.org/glossary/detail.asp?ID=5408
    The numbers in those categories are significant for some countries, but not for Mexico.
    I am not sure exactly which of those two categories the IAB includes in "highskilled."

    > 92k is the number who wrote the entrance exam
    That is why I qualified that with “unless it is in a very high IQ region”.
     
    I doubt that even in a Mexican (or American) high IQ region EVERYONE graduates high school and applies to college. Did you miss my point?

    Any thoughts on the age group issue I raised based on Table A1.2?


    Dispite what the American thinks of the Mexican immigrants, for UK it seems to be a different picture due to the selection policies,
     
    As you observe, the US and UK have a very different profile for Mexican immigrants. Here are the IAB 2010 low/medium/high skill numbers for Mexican immigrants:
    US Men: 2900516 1468257 675837
    GBR Men: 188 115 2390
    so in the US the low/high ratio is 4.3 while in the UK the ratio is 0.08. Given that (a factor of 50 different!), of course the picture is different. The size of the groups involved (a factor of 2000!) also makes a difference.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  38. res says:
    @dux.ie
    ENIGMA is a first order coarse grain estimation simply working off the bell curve mathematics and I need that to estimate the migrant average group IQ (MAGIQ) rather than plucking numbers from thin air. The implicit assumptions are that there are complete meritocracy and frictionless university student selection. If the deviations from the assumptions are known then ENIGMA could be modified to accomodate that. For example India seems to have formal quota for the minorities but I do not have those data.

    Immigrant group data like that from IAB usually have the data on the fraction that have university degrees. Pivoting on that and with the ENIGMA value and doing the bell curve maths again can give a ball park figure for the immigrant group IQ. For example the Becker IQ for India is 83.23 and OECD data for the national fraction of gradute is 5.64% that give ENIGMA of 107.02, and the fraction of Indian immigrants to UK with degree (FracHi) from IAB is 45.58% that gives the Indian migrant group IQ to UK MAGIQ=105.35, a bit higher than the CAT estimation of the IQ of the children of Indian immigrants but much higher than the source country national average IQ. May be I should have factored in the effects of the Indian quota system. The effects of the migrant selection policy are more evident for the Nigerian with IQdb=70.59, from UNICEF UniA=7%, from IAB FracHi=72.28% (!), giving MAGIC=101.6, a jump of 30.1 IQ points. For HKG the IQdb for China is used as proxy and the value could be a bit too high.

    Host UK
    Rank IQcat MAGIQ ENIGMA IQdb UniA FracHi Country
    23 99.8 105.35 107.02 83.23 5.64 45.58 IND
    29 93.4 101.6 92.73 70.59 7.0 72.28(!) NGA Nigeria
    34 93.5 97.27 105.53 79.94 4.4 29.1 PAK
    37 94.7 93.37 104.49 74.56 2.3 22.92 BGD
    42 94.0 85.46 94.28 73.6 8.4 27.82 JAM
    6 107.5 117.21 120.46 106.27 17.21 41.42 HKG

    The Table A1.1 you referred to has zero decimal places and the rounding off error could be significant for low percentages. An 2015 OECD report which I cannot locate now has at least one decimal place data and the values seem to be larger. Some diff

    EAG17|EAG15?|DIF|Country
    31|44.0|-13.0|AUS
    16|20.7|-4.7|AUT
    37|39.1|-2.1|BEL
    22|33.8|-11.8|CZE
    32|49.8|-17.8|DNK

    Furthermore the data on OECD partners cannot be trusted. For example the data for China at 3% (which gives ENIGMA=133.25) has not been updated for quite some time, the EAG15?? listed 26.4% (ENIGMA=115) seems to be more reasonable.

    > 92k is the number who wrote the entrance exam
    That is why I qualified that with "unless it is in a very high IQ region". Actually the 0.251 number was not that high, a more recent UniA seems to be .238 compares to the old 0.162 , that put the new ENIGMA=99.2 even though IQdb added 0.51 IQ point to that from IQLynn. Dispite what the American thinks of the Mexican immigrants, for UK it seems to be a different picture due to the selection policies,

    Rank MAGIQ ENIGMA IQdb UniA FracHi Country
    9 113.74 99.2 88.51 23.8 83.38(!!) MEX

    Immigrant group data like that from IAB usually have the data on the fraction that have university degrees.

    My understanding is that the IAB Brain Drain data uses “some tertiary education” rather than university degree. From their methodological note at http://www.iab.de/en/daten/iab-brain-drain-data.aspx

    Educational categories. We distinguish three levels of education: primary (low skilled: includes lower secondary, primary and no schooling); secondary (mediumskilled: high-school leaving certificate or equivalent) and tertiary education (highskilled: higher than high-school leaving certificate or equivalent).

    If I understand correctly that is equivalent to “some college.” Do you know differently? Table A1.1 has the categories “Post-secondary non-tertiary” and “Short-cycle tertiary.” Here is some documentation for what those terms mean: https://www.oecd-ilibrary.org/education/isced-2011-operational-manual/isced-2011-level-5-short-cycle-tertiary-education_9789264228368-10-en https://stats.oecd.org/glossary/detail.asp?ID=5408
    The numbers in those categories are significant for some countries, but not for Mexico.
    I am not sure exactly which of those two categories the IAB includes in “highskilled.”

    > 92k is the number who wrote the entrance exam
    That is why I qualified that with “unless it is in a very high IQ region”.

    I doubt that even in a Mexican (or American) high IQ region EVERYONE graduates high school and applies to college. Did you miss my point?

    Any thoughts on the age group issue I raised based on Table A1.2?

    Dispite what the American thinks of the Mexican immigrants, for UK it seems to be a different picture due to the selection policies,

    As you observe, the US and UK have a very different profile for Mexican immigrants. Here are the IAB 2010 low/medium/high skill numbers for Mexican immigrants:
    US Men: 2900516 1468257 675837
    GBR Men: 188 115 2390
    so in the US the low/high ratio is 4.3 while in the UK the ratio is 0.08. Given that (a factor of 50 different!), of course the picture is different. The size of the groups involved (a factor of 2000!) also makes a difference.

    Read More
    • Replies: @James Thompson
    Interesting work @dux.ie and @res, and thanks for telling us about it. Couple of problems. India is very variable in ability. Provinces vary enormously, and also by castes and clans, the latter in ways which may be hard to track down without Indian specialists to explain them. PISA helps, but countries tend to cheat a bit. Mingrui Wang has data which suggests estimates for Chinese intelligence may be inflated by using almost exclusively urban samples. Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai (this from memory, but certainly a small sample which was university based).
    So, I like the method of limits, but countries differ in how honestly and consistently they grant university entrance. None of this is fatal to estimation, but it all causes some anomalies.
    Also, UK immigration can be weird. Absolute jumble of abilities, as you would expect if they were not selecting for ability.
    , @dux.ie
    For those countries of interest like Jamaica and Nigeria, you'll be lucky to get a single %Grad from UNESCO, in Nigeria's case it was even the average %Grad for the Sub-Saharan region, and you worried about if the post-secondory training data is included?? Anyway, I have 4 versions of OECD data eag15s, eag15t, eag15n and eag15z the sources of which were in thumb drives somewhere else. Example of a version I did not used (eag15s),

    PostSec|UniB|UniA|Adv|Sum|Country
    4.37663263125|10.4369802772|27.1221465197|0.78294424121|42.71870366936|AUS

    Compare to the data you cited,

    PostSec|ShortTer|Bachelor|Master|Phd|Country
    5|11|24|6|1|AUS

    so the Master values seems to be absorbed into UniA, the summed value of 42.71870366936% is close to the 44.00% value I used for AUS from eag15n which had separate %Grad (not sure with respect to which pop base) for oversea students and thus might give some idea the likelyhood that some of the immigrants had host country university degrees. Thus the so called 'UniA' values I labelled is the sum of the whole lot if they are available. Note the number of decimal places available for the numbers. I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve. So what is your problem?? There are a few engineers who refuse to fly in airplanes because they are designed with imprecise floating point numbers. I hope you are not an aerospace engineer.

    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  39. @dux.ie
    ENIGMA is just a fancy name for simple bell curve mathematics so far only a first order coarse grain estimation based on average national IQ and the fraction of population with university degrees. Refinement is possible but age segmented IQ data are generally not available. No time to elaborate now. Frac115 is just given the average national IQ the fraction of the population with IQ ≥ 115.

    Look, I’ll be the first to admit that I have a very disorganized academic past culminating in a US BSBA from a second tier state supported university (that still had a borderline first tier B-School in the mid-seventies) that I never really needed in the absolute technical academic proficiency sense of the word. I did learn to use a slide rule quite proficiently in a calculus course taught by an Economics professor and used same quite proficiently in Banking and Finance classes. Made a D in Statistics taught by a Greek (I don’t mean a frat rat either) whose spoken English was so bad, he disappeared after one semester. Repeated course with a B under Chairman of Economics Dept. All of this was occurring as the transition from slide rule to H-P calculators was looming up front and close and computers were still a main frame a hundred miles away and FORTRAN uploads from punched card stacks were transmitted late at night by grad students when AT&T long distance phone rates were (relatively) low.

    So going back to my question (19) about Mexico which generated the response from dux.ie (21) with the chart a few days back, I’m still scratchin’ my little low IQ Columbo head about a couple of items after spending a fair amount of time just decompressing the chart’s columns and lining them up with their respective headings:

    RE: country names — AUT = Austria or Australia?? DEU= Germany ??

    RE: number row for USA– How is it that 30.8% of the population of the USA has some sort of college degree, when only 12.9% of the population is admitted to a college to start with? Surely something has flown over my head here.

    Read More
    • Replies: @res

    RE: country names — AUT = Austria or Australia?? DEU= Germany ??
     
    Three letter country codes are standardized (assuming dux.ie is using those): http://www.nationsonline.org/oneworld/country_code_list.htm
    AUT is Austria (Australia is AUS), DEU is Germany

    RE: number row for USA– How is it that 30.8% of the population of the USA has some sort of college degree, when only 12.9% of the population is admitted to a college to start with? Surely something has flown over my head here.
     
    12.9% is the US fraction (estimated) with IQ over 115 (not admitted to college).
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  40. res says:
    @Jim Bob Lassiter
    Look, I'll be the first to admit that I have a very disorganized academic past culminating in a US BSBA from a second tier state supported university (that still had a borderline first tier B-School in the mid-seventies) that I never really needed in the absolute technical academic proficiency sense of the word. I did learn to use a slide rule quite proficiently in a calculus course taught by an Economics professor and used same quite proficiently in Banking and Finance classes. Made a D in Statistics taught by a Greek (I don't mean a frat rat either) whose spoken English was so bad, he disappeared after one semester. Repeated course with a B under Chairman of Economics Dept. All of this was occurring as the transition from slide rule to H-P calculators was looming up front and close and computers were still a main frame a hundred miles away and FORTRAN uploads from punched card stacks were transmitted late at night by grad students when AT&T long distance phone rates were (relatively) low.

    So going back to my question (19) about Mexico which generated the response from dux.ie (21) with the chart a few days back, I'm still scratchin' my little low IQ Columbo head about a couple of items after spending a fair amount of time just decompressing the chart's columns and lining them up with their respective headings:

    RE: country names -- AUT = Austria or Australia?? DEU= Germany ??

    RE: number row for USA-- How is it that 30.8% of the population of the USA has some sort of college degree, when only 12.9% of the population is admitted to a college to start with? Surely something has flown over my head here.

    RE: country names — AUT = Austria or Australia?? DEU= Germany ??

    Three letter country codes are standardized (assuming dux.ie is using those): http://www.nationsonline.org/oneworld/country_code_list.htm
    AUT is Austria (Australia is AUS), DEU is Germany

    RE: number row for USA– How is it that 30.8% of the population of the USA has some sort of college degree, when only 12.9% of the population is admitted to a college to start with? Surely something has flown over my head here.

    12.9% is the US fraction (estimated) with IQ over 115 (not admitted to college).

    Read More
    • Replies: @Jim Bob Lassiter
    I accidentally set up my reply to you as stand alone comment #45. Please accept my apologies. Been having some Mr. McGoo moments of late.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  41. @res

    Immigrant group data like that from IAB usually have the data on the fraction that have university degrees.
     
    My understanding is that the IAB Brain Drain data uses "some tertiary education" rather than university degree. From their methodological note at http://www.iab.de/en/daten/iab-brain-drain-data.aspx

    Educational categories. We distinguish three levels of education: primary (low skilled: includes lower secondary, primary and no schooling); secondary (mediumskilled: high-school leaving certificate or equivalent) and tertiary education (highskilled: higher than high-school leaving certificate or equivalent).
     
    If I understand correctly that is equivalent to "some college." Do you know differently? Table A1.1 has the categories "Post-secondary non-tertiary" and "Short-cycle tertiary." Here is some documentation for what those terms mean: https://www.oecd-ilibrary.org/education/isced-2011-operational-manual/isced-2011-level-5-short-cycle-tertiary-education_9789264228368-10-en https://stats.oecd.org/glossary/detail.asp?ID=5408
    The numbers in those categories are significant for some countries, but not for Mexico.
    I am not sure exactly which of those two categories the IAB includes in "highskilled."

    > 92k is the number who wrote the entrance exam
    That is why I qualified that with “unless it is in a very high IQ region”.
     
    I doubt that even in a Mexican (or American) high IQ region EVERYONE graduates high school and applies to college. Did you miss my point?

    Any thoughts on the age group issue I raised based on Table A1.2?


    Dispite what the American thinks of the Mexican immigrants, for UK it seems to be a different picture due to the selection policies,
     
    As you observe, the US and UK have a very different profile for Mexican immigrants. Here are the IAB 2010 low/medium/high skill numbers for Mexican immigrants:
    US Men: 2900516 1468257 675837
    GBR Men: 188 115 2390
    so in the US the low/high ratio is 4.3 while in the UK the ratio is 0.08. Given that (a factor of 50 different!), of course the picture is different. The size of the groups involved (a factor of 2000!) also makes a difference.

    Interesting work and , and thanks for telling us about it. Couple of problems. India is very variable in ability. Provinces vary enormously, and also by castes and clans, the latter in ways which may be hard to track down without Indian specialists to explain them. PISA helps, but countries tend to cheat a bit. Mingrui Wang has data which suggests estimates for Chinese intelligence may be inflated by using almost exclusively urban samples. Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai (this from memory, but certainly a small sample which was university based).
    So, I like the method of limits, but countries differ in how honestly and consistently they grant university entrance. None of this is fatal to estimation, but it all causes some anomalies.
    Also, UK immigration can be weird. Absolute jumble of abilities, as you would expect if they were not selecting for ability.

    Read More
    • Replies: @res
    Thanks for your kind words, but I actually learned about the Brain Drain data from Emil in a comment (a year ago today ; ) at your blog: http://www.unz.com/jthompson/immigration-is-it-the-people/#comment-1893308
    So the credit goes to the two of you.

    Couple of problems. India is very variable in ability.
     
    Good point. Subpopulations can be a major problem--especially for approaches which focus on the tails. Even more complicated when the subpopulations vary greatly in all of nature, nurture, and phenotype.

    So, I like the method of limits, but countries differ in how honestly and consistently they grant university entrance. None of this is fatal to estimation, but it all causes some anomalies.
     
    Agreed.

    Also, UK immigration can be weird. Absolute jumble of abilities, as you would expect if they were not selecting for ability.
     
    Understood. The US can be the same way. It is interesting how often a pattern appears of the medium skilled being underrepresented compared to the low and high skilled. I have been assuming this is due to a combination of effects favoring both refugees and high ability immigrants.

    Regarding immigration selectivity, I think it is important to differentiate between policy based and immigrant choice/ability based selectivity (Chanda Chisala appears utterly unable to understand this point). One of the reason I like the IAB data is it provides an empirical measure of selectivity. The why and how of that data are non-trivial questions.
    , @dux.ie
    > Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai

    There are always people whinging about the PISA 2012 Shanghai results being from a urban area and needed to scale down for China. The Piffer's results of the alleged Shanghai University graduates' genomes under-predicting the Chinese intelligence might be closer to the truth.

    First, it is not always true that city students perform better than the rural students. The American ivy league professor without any proof or data raised that question when the US NAEP test results showed the rural was doing better than city, town and suburban, table 6b. https://nces.ed.gov/programs/coe/pdf/coe_tla.pdf Are there any Chinese data? Recently the Chinese Education Dept issued an directive against any sources highlighting the high performance students in the College Entrance Exams (CEE GaoKao). So there will be no new data and the existing data will be progressively removed. Thus try you luck on this photo,
    https://www.whatsonweibo.com/wp-content/uploads/2018/05/917aaba497edfe6e-422x360.jpg
    the top two students were the top scorers for the CEE science (693) and arts (629) from the counties while the bottom two were that from the cities (respectively 685, 613). Both the counties students beat the city students in their respective majors. The four mentioned students were the top students from rural counties and cities among the about 10 million students sat the CEE. By sheer number out of 10 million students they can afford to select only 0.03% (IQlike=178) to be admitted to the top 2 Peking and Tsinghua Universities.

    Are there any more global data beside anecdotal evidence? Dr. Hongbin Li, James Liang Director of the China Program, Stanford Center on Global Poverty and Development, Stanford University, showed the statistics of the Chinese CEE for all students in the urban and rural areas, https://www.youtube.com/watch?v=lP3QW8OgVJE The data at 16:45 the screen capture,
    http://tinypic.com/view.php?pic=2ilhi83&s=9
    The results for the urban students seem to agree to the assumption that most of them are descendents of the ancient Chinese imperial court officials who passed the imperial court exams writing the "eight legged essays". They greatly excelled in languages like Mandarin and English whereas most of those from the rural areas greatly excelled in mathematics. There is no centralized CEE and each administrative regions set their own exams and the level of the maximum scores. Thus the national average is the average of the provincial average with respect to their maximum scores. However I still do not understand why the sum of the rural and city averages come out to 1. Anyway, for the total scores and the composite scores the rural students performed better than the urban students on a national scale. Thus as Piffer found out, if the Chinese PISA score is to be modified it might have to move upwards.

    As to the quality of the Shanghai university students, that depends on the CEE scores distribution and the number of available local universities. The governance of the Chinese administrative regions is pretty autonomous in certain areas like education. Thus rich regions can finance more universities which have preference for local students with the result that the minimum entrance scores will be reduced. Shanghai being a rich region no doubt can attract many top students. However, with many universities there, the resulting average student quality is not that good. From a ranked list of the cut-off percentage off the respective max provincial scores ratio1, Shanghai ranked third from the bottom, or weighted by the simple ratio of IQ over the national average IQ, W1, rank 16 just below the middle.

    Rank|W1|ratio1|RegionE
    1|86.2|76.9|Zhejiang
    2|81.7|74.0|Beijing
    3|73.7|70.9|Liaoning
    16|67.9|60.9|Shanghai
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  42. dux.ie says:
    @res

    Immigrant group data like that from IAB usually have the data on the fraction that have university degrees.
     
    My understanding is that the IAB Brain Drain data uses "some tertiary education" rather than university degree. From their methodological note at http://www.iab.de/en/daten/iab-brain-drain-data.aspx

    Educational categories. We distinguish three levels of education: primary (low skilled: includes lower secondary, primary and no schooling); secondary (mediumskilled: high-school leaving certificate or equivalent) and tertiary education (highskilled: higher than high-school leaving certificate or equivalent).
     
    If I understand correctly that is equivalent to "some college." Do you know differently? Table A1.1 has the categories "Post-secondary non-tertiary" and "Short-cycle tertiary." Here is some documentation for what those terms mean: https://www.oecd-ilibrary.org/education/isced-2011-operational-manual/isced-2011-level-5-short-cycle-tertiary-education_9789264228368-10-en https://stats.oecd.org/glossary/detail.asp?ID=5408
    The numbers in those categories are significant for some countries, but not for Mexico.
    I am not sure exactly which of those two categories the IAB includes in "highskilled."

    > 92k is the number who wrote the entrance exam
    That is why I qualified that with “unless it is in a very high IQ region”.
     
    I doubt that even in a Mexican (or American) high IQ region EVERYONE graduates high school and applies to college. Did you miss my point?

    Any thoughts on the age group issue I raised based on Table A1.2?


    Dispite what the American thinks of the Mexican immigrants, for UK it seems to be a different picture due to the selection policies,
     
    As you observe, the US and UK have a very different profile for Mexican immigrants. Here are the IAB 2010 low/medium/high skill numbers for Mexican immigrants:
    US Men: 2900516 1468257 675837
    GBR Men: 188 115 2390
    so in the US the low/high ratio is 4.3 while in the UK the ratio is 0.08. Given that (a factor of 50 different!), of course the picture is different. The size of the groups involved (a factor of 2000!) also makes a difference.

    For those countries of interest like Jamaica and Nigeria, you’ll be lucky to get a single %Grad from UNESCO, in Nigeria’s case it was even the average %Grad for the Sub-Saharan region, and you worried about if the post-secondory training data is included?? Anyway, I have 4 versions of OECD data eag15s, eag15t, eag15n and eag15z the sources of which were in thumb drives somewhere else. Example of a version I did not used (eag15s),

    PostSec|UniB|UniA|Adv|Sum|Country
    4.37663263125|10.4369802772|27.1221465197|0.78294424121|42.71870366936|AUS

    Compare to the data you cited,

    PostSec|ShortTer|Bachelor|Master|Phd|Country
    5|11|24|6|1|AUS

    so the Master values seems to be absorbed into UniA, the summed value of 42.71870366936% is close to the 44.00% value I used for AUS from eag15n which had separate %Grad (not sure with respect to which pop base) for oversea students and thus might give some idea the likelyhood that some of the immigrants had host country university degrees. Thus the so called ‘UniA’ values I labelled is the sum of the whole lot if they are available. Note the number of decimal places available for the numbers. I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve. So what is your problem?? There are a few engineers who refuse to fly in airplanes because they are designed with imprecise floating point numbers. I hope you are not an aerospace engineer.

    Read More
    • Replies: @res
    It would be great if you could provide a reference so the rest of us can look at the same higher accuracy data you are using.

    For those countries of interest like Jamaica and Nigeria, you’ll be lucky to get a single %Grad from UNESCO, in Nigeria’s case it was even the average %Grad for the Sub-Saharan region, and you worried about if the post-secondory training data is included??
     
    That is a worthwhile point about countries like Nigeria. Then there are countries like Canada where about half of the tertiary education is in the form of short cycle tertiary (per Table A1.1 above). I think even aerospace engineers care about a factor of two.

    I have 4 versions of OECD data eag15s, eag15t, eag15n and eag15z
     
    That's nice. I assume s and t are secondary and tertiary. What are n and z? Giving me a barrage of cryptic variable names unaccompanied by further explanation or the actual data (complete, with source) is not terribly persuasive.

    Thus the so called ‘UniA’ values I labelled is the sum of the whole lot if they are available.
     
    Thank you for clarifying. Can I assume you accounted for the d and x (data grouped in other categories) issue I raised in the PS to comment 23? (there was a reason I included that PS. It is mostly an issue for the partners, but also for some of the OECD countries)

    There are a few engineers who refuse to fly in airplanes because they are designed with imprecise floating point numbers. I hope you are not an aerospace engineer.
     
    In response to that I will just note the end of my comment 23 where I raise a factor of three issue with your analysis and you just wave it away. Are you an aerospace engineer? Have you ever had to care about floating point precision professionally? (I have)

    P.S. One clue that you are not an engineer-quoting 12 significant digits for numbers that would be lucky to be accurate to 3. If you are depending on that precision you are a fool.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  43. res says:
    @dux.ie
    For those countries of interest like Jamaica and Nigeria, you'll be lucky to get a single %Grad from UNESCO, in Nigeria's case it was even the average %Grad for the Sub-Saharan region, and you worried about if the post-secondory training data is included?? Anyway, I have 4 versions of OECD data eag15s, eag15t, eag15n and eag15z the sources of which were in thumb drives somewhere else. Example of a version I did not used (eag15s),

    PostSec|UniB|UniA|Adv|Sum|Country
    4.37663263125|10.4369802772|27.1221465197|0.78294424121|42.71870366936|AUS

    Compare to the data you cited,

    PostSec|ShortTer|Bachelor|Master|Phd|Country
    5|11|24|6|1|AUS

    so the Master values seems to be absorbed into UniA, the summed value of 42.71870366936% is close to the 44.00% value I used for AUS from eag15n which had separate %Grad (not sure with respect to which pop base) for oversea students and thus might give some idea the likelyhood that some of the immigrants had host country university degrees. Thus the so called 'UniA' values I labelled is the sum of the whole lot if they are available. Note the number of decimal places available for the numbers. I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve. So what is your problem?? There are a few engineers who refuse to fly in airplanes because they are designed with imprecise floating point numbers. I hope you are not an aerospace engineer.

    It would be great if you could provide a reference so the rest of us can look at the same higher accuracy data you are using.

    For those countries of interest like Jamaica and Nigeria, you’ll be lucky to get a single %Grad from UNESCO, in Nigeria’s case it was even the average %Grad for the Sub-Saharan region, and you worried about if the post-secondory training data is included??

    That is a worthwhile point about countries like Nigeria. Then there are countries like Canada where about half of the tertiary education is in the form of short cycle tertiary (per Table A1.1 above). I think even aerospace engineers care about a factor of two.

    I have 4 versions of OECD data eag15s, eag15t, eag15n and eag15z

    That’s nice. I assume s and t are secondary and tertiary. What are n and z? Giving me a barrage of cryptic variable names unaccompanied by further explanation or the actual data (complete, with source) is not terribly persuasive.

    Thus the so called ‘UniA’ values I labelled is the sum of the whole lot if they are available.

    Thank you for clarifying. Can I assume you accounted for the d and x (data grouped in other categories) issue I raised in the PS to comment 23? (there was a reason I included that PS. It is mostly an issue for the partners, but also for some of the OECD countries)

    There are a few engineers who refuse to fly in airplanes because they are designed with imprecise floating point numbers. I hope you are not an aerospace engineer.

    In response to that I will just note the end of my comment 23 where I raise a factor of three issue with your analysis and you just wave it away. Are you an aerospace engineer? Have you ever had to care about floating point precision professionally? (I have)

    P.S. One clue that you are not an engineer-quoting 12 significant digits for numbers that would be lucky to be accurate to 3. If you are depending on that precision you are a fool.

    Read More
    • Replies: @dux.ie
    > I actually learned about the Brain Drain data from Emil in a comment (a year ago today ; )

    I mentioned the IAB data in this blog back in 2016, comments 7,8,9. Dr. Thompson asked for the source.
    https://www.unz.com/jthompson/the-accuracy-of-stereotypes/
    That was when I did the analysis. I've fuzzy memory about them now.

    >I am not sure exactly which of those two categories the IAB includes in “highskilled.”

    IAB aggregated the census data from host countries. Canada is transparent about their immigrant selection process. In the Canadian process and I assume other countries' as well, 3-year diplomas are lumped together with university degrees,

    https://www.canada.ca/en/immigration-refugees-citizenship/services/immigrate-canada/express-entry/become-candidate/eligibility/education-assessed/read-report.html

    """University degree at the Doctoral (PhD) level 140 points
    University degree at the Master’s level or Licensed Professional degree 126 points"""

    """Bachelor's degree or other programs (three or more years) at a university, college, trade or technical school, or other institute 112 points
    ECA assessment results:
    College Diploma (three years)
    Diploma (three years)
    Three-year diploma in [name of discipline]
    Three-year advanced diploma
    Bachelor's Degree ... """

    Even the two years diplomas or extended high school diplomas are not that far behind, still categorized them as two-year degree, this could be the “Short-cycle tertiary”

    """Two-year degree, diploma or certificate from a program at a university, college, trade or technical school, or other institute 91 points
    ECA assessment results:
    College Diploma
    College Diploma (two years)
    Diploma (two years)
    Two-year diploma
    Two-year diploma in [name of discipline]
    Secondary school diploma and diploma (two years)
    Associate Degree
    Associate of [Arts/Science] degree
    Post-secondary Diploma with a focus in [area of concentration]
    Two-year post-secondary Diploma with a focus in [area of concentration]
    Two-year associate degree"""

    This could be the “Post-secondary non-tertiary”

    """One-year degree, diploma or certificate from a program at a university, college, trade or technical school, or other institute 84 points
    ECA assessment results:
    Certificate ... """

    > It would be great if you could provide a reference

    That was from a file most probably named "EAG_GRAD_ENTR_RATES_09022018023322785" (unlikely to be the name I would type in). Google search comes up empty. The data has values for W1 which they termed "Mobile Students" and I assumed that to be oversea students, and W0 is "World (all entities, including reference area, including IO)" I assumed to be the total number of graduates from that country dispite the word "World". Google seach comes up with some hits for that term. None of them from OECD. Seems to be a standard EU category. The values of W0-W1 listed is similar to that of table A1.1. That was quite some time ago, my memory is a bit fuzzy about that. I only saved a csv copy. I have a few thumb drives bit rot in the last 2 years. Beware of them as you will only realize that when you open them up and they are full of nulls while the file sizes still remain the same. Backing up of null files are pretty useless.

    > I assume s and t are secondary and tertiary. What are n and z?

    For some forgotten reason I suffixed the first as 's', thus the next one was naturally 't'. And the newer one 'n'. 'z' is the last I would bother with from that particular source and year, it seems to be similar to the Table A1.1 with no decimal places.

    > d and x

    Missing values. Cannot filled in from external sources. These UniA values are from UNESCO or country census,

    ISO|Country|UniA
    BGD|Bangladesh|2.3
    HKG|HongKong|17.21
    IND|India|5.64
    IRN|Iran|18.53
    JAM|Jamaica|8.4
    NGA|Nigeria|7.

    Look. Comment section can only talk about overall methodology and that was already too long, not detailed calculations or citing every individual sources. Even on methodology I already said too much about the initial assumptions. Elsewhere the discussions had been bogged down on the assumption of frictionless uni student selections. The simplifying assumptions are there to be able to arrived at some ball park figures or just simply throwing up your hands and gives up. The immigrants are already in the country and it is whether you want to know about their approximate abilities. Some just simply want to shut down the discussions.

    > Have you ever had to care about floating point precision professionally? (I have)

    Did you missed it? I said "I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve." For example, many of Pew research data with small percentages less than 5 had no decimal places which rendered them useless for further exploration. I know you can request raw data but I prefer not to leave too many digital footprint.

    > P.S. One clue that you are not an engineer-quoting 12 significant digits for numbers that would be lucky to be accurate to 3. If you are depending on that precision you are a fool.

    You don't seem to understand the difference between highlighting the RAW data used in the context of the discussion from the final results presentation. You seems to be pretty an*l.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  44. res says:
    @James Thompson
    Interesting work @dux.ie and @res, and thanks for telling us about it. Couple of problems. India is very variable in ability. Provinces vary enormously, and also by castes and clans, the latter in ways which may be hard to track down without Indian specialists to explain them. PISA helps, but countries tend to cheat a bit. Mingrui Wang has data which suggests estimates for Chinese intelligence may be inflated by using almost exclusively urban samples. Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai (this from memory, but certainly a small sample which was university based).
    So, I like the method of limits, but countries differ in how honestly and consistently they grant university entrance. None of this is fatal to estimation, but it all causes some anomalies.
    Also, UK immigration can be weird. Absolute jumble of abilities, as you would expect if they were not selecting for ability.

    Thanks for your kind words, but I actually learned about the Brain Drain data from Emil in a comment (a year ago today ; ) at your blog: http://www.unz.com/jthompson/immigration-is-it-the-people/#comment-1893308
    So the credit goes to the two of you.

    Couple of problems. India is very variable in ability.

    Good point. Subpopulations can be a major problem–especially for approaches which focus on the tails. Even more complicated when the subpopulations vary greatly in all of nature, nurture, and phenotype.

    So, I like the method of limits, but countries differ in how honestly and consistently they grant university entrance. None of this is fatal to estimation, but it all causes some anomalies.

    Agreed.

    Also, UK immigration can be weird. Absolute jumble of abilities, as you would expect if they were not selecting for ability.

    Understood. The US can be the same way. It is interesting how often a pattern appears of the medium skilled being underrepresented compared to the low and high skilled. I have been assuming this is due to a combination of effects favoring both refugees and high ability immigrants.

    Regarding immigration selectivity, I think it is important to differentiate between policy based and immigrant choice/ability based selectivity (Chanda Chisala appears utterly unable to understand this point). One of the reason I like the IAB data is it provides an empirical measure of selectivity. The why and how of that data are non-trivial questions.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  45. Thank you very much. That column heading definition correction flicked the switch (or pulled the chain; I forget which) on my Ford light bulb. I can instantly intuit some correlations and causations just by looking at my re-written chart. No tedious higher math calculations, no pedantry necessary.

    Now if I can just memorize all these new-fangled international country abbreviations, I’ll truly be a Reformed New Global Citizen.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  46. @res

    RE: country names — AUT = Austria or Australia?? DEU= Germany ??
     
    Three letter country codes are standardized (assuming dux.ie is using those): http://www.nationsonline.org/oneworld/country_code_list.htm
    AUT is Austria (Australia is AUS), DEU is Germany

    RE: number row for USA– How is it that 30.8% of the population of the USA has some sort of college degree, when only 12.9% of the population is admitted to a college to start with? Surely something has flown over my head here.
     
    12.9% is the US fraction (estimated) with IQ over 115 (not admitted to college).

    I accidentally set up my reply to you as stand alone comment #45. Please accept my apologies. Been having some Mr. McGoo moments of late.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  47. dux.ie says:
    @res
    It would be great if you could provide a reference so the rest of us can look at the same higher accuracy data you are using.

    For those countries of interest like Jamaica and Nigeria, you’ll be lucky to get a single %Grad from UNESCO, in Nigeria’s case it was even the average %Grad for the Sub-Saharan region, and you worried about if the post-secondory training data is included??
     
    That is a worthwhile point about countries like Nigeria. Then there are countries like Canada where about half of the tertiary education is in the form of short cycle tertiary (per Table A1.1 above). I think even aerospace engineers care about a factor of two.

    I have 4 versions of OECD data eag15s, eag15t, eag15n and eag15z
     
    That's nice. I assume s and t are secondary and tertiary. What are n and z? Giving me a barrage of cryptic variable names unaccompanied by further explanation or the actual data (complete, with source) is not terribly persuasive.

    Thus the so called ‘UniA’ values I labelled is the sum of the whole lot if they are available.
     
    Thank you for clarifying. Can I assume you accounted for the d and x (data grouped in other categories) issue I raised in the PS to comment 23? (there was a reason I included that PS. It is mostly an issue for the partners, but also for some of the OECD countries)

    There are a few engineers who refuse to fly in airplanes because they are designed with imprecise floating point numbers. I hope you are not an aerospace engineer.
     
    In response to that I will just note the end of my comment 23 where I raise a factor of three issue with your analysis and you just wave it away. Are you an aerospace engineer? Have you ever had to care about floating point precision professionally? (I have)

    P.S. One clue that you are not an engineer-quoting 12 significant digits for numbers that would be lucky to be accurate to 3. If you are depending on that precision you are a fool.

    > I actually learned about the Brain Drain data from Emil in a comment (a year ago today ; )

    I mentioned the IAB data in this blog back in 2016, comments 7,8,9. Dr. Thompson asked for the source.

    https://www.unz.com/jthompson/the-accuracy-of-stereotypes/

    That was when I did the analysis. I’ve fuzzy memory about them now.

    >I am not sure exactly which of those two categories the IAB includes in “highskilled.”

    IAB aggregated the census data from host countries. Canada is transparent about their immigrant selection process. In the Canadian process and I assume other countries’ as well, 3-year diplomas are lumped together with university degrees,

    https://www.canada.ca/en/immigration-refugees-citizenship/services/immigrate-canada/express-entry/become-candidate/eligibility/education-assessed/read-report.html

    “””University degree at the Doctoral (PhD) level 140 points
    University degree at the Master’s level or Licensed Professional degree 126 points”””

    “””Bachelor’s degree or other programs (three or more years) at a university, college, trade or technical school, or other institute 112 points
    ECA assessment results:
    College Diploma (three years)
    Diploma (three years)
    Three-year diploma in [name of discipline]
    Three-year advanced diploma
    Bachelor’s Degree … “””

    Even the two years diplomas or extended high school diplomas are not that far behind, still categorized them as two-year degree, this could be the “Short-cycle tertiary”

    “””Two-year degree, diploma or certificate from a program at a university, college, trade or technical school, or other institute 91 points
    ECA assessment results:
    College Diploma
    College Diploma (two years)
    Diploma (two years)
    Two-year diploma
    Two-year diploma in [name of discipline]
    Secondary school diploma and diploma (two years)
    Associate Degree
    Associate of [Arts/Science] degree
    Post-secondary Diploma with a focus in [area of concentration]
    Two-year post-secondary Diploma with a focus in [area of concentration]
    Two-year associate degree”””

    This could be the “Post-secondary non-tertiary”

    “””One-year degree, diploma or certificate from a program at a university, college, trade or technical school, or other institute 84 points
    ECA assessment results:
    Certificate … “””

    > It would be great if you could provide a reference

    That was from a file most probably named “EAG_GRAD_ENTR_RATES_09022018023322785″ (unlikely to be the name I would type in). Google search comes up empty. The data has values for W1 which they termed “Mobile Students” and I assumed that to be oversea students, and W0 is “World (all entities, including reference area, including IO)” I assumed to be the total number of graduates from that country dispite the word “World”. Google seach comes up with some hits for that term. None of them from OECD. Seems to be a standard EU category. The values of W0-W1 listed is similar to that of table A1.1. That was quite some time ago, my memory is a bit fuzzy about that. I only saved a csv copy. I have a few thumb drives bit rot in the last 2 years. Beware of them as you will only realize that when you open them up and they are full of nulls while the file sizes still remain the same. Backing up of null files are pretty useless.

    > I assume s and t are secondary and tertiary. What are n and z?

    For some forgotten reason I suffixed the first as ‘s’, thus the next one was naturally ‘t’. And the newer one ‘n’. ‘z’ is the last I would bother with from that particular source and year, it seems to be similar to the Table A1.1 with no decimal places.

    > d and x

    Missing values. Cannot filled in from external sources. These UniA values are from UNESCO or country census,

    ISO|Country|UniA
    BGD|Bangladesh|2.3
    HKG|HongKong|17.21
    IND|India|5.64
    IRN|Iran|18.53
    JAM|Jamaica|8.4
    NGA|Nigeria|7.

    Look. Comment section can only talk about overall methodology and that was already too long, not detailed calculations or citing every individual sources. Even on methodology I already said too much about the initial assumptions. Elsewhere the discussions had been bogged down on the assumption of frictionless uni student selections. The simplifying assumptions are there to be able to arrived at some ball park figures or just simply throwing up your hands and gives up. The immigrants are already in the country and it is whether you want to know about their approximate abilities. Some just simply want to shut down the discussions.

    > Have you ever had to care about floating point precision professionally? (I have)

    Did you missed it? I said “I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve.” For example, many of Pew research data with small percentages less than 5 had no decimal places which rendered them useless for further exploration. I know you can request raw data but I prefer not to leave too many digital footprint.

    > P.S. One clue that you are not an engineer-quoting 12 significant digits for numbers that would be lucky to be accurate to 3. If you are depending on that precision you are a fool.

    You don’t seem to understand the difference between highlighting the RAW data used in the context of the discussion from the final results presentation. You seems to be pretty an*l.

    Read More
    • Replies: @res

    I mentioned the IAB data in this blog back in 2016, comments 7,8,9. Dr. Thompson asked for the source.
     
    Sorry that I missed that. Sometimes learning about something requires having a need for it, and the IAB data was very useful in Chanda Chisala threads that happened in the interim.

    On reflection, I notice that that post was one of the last before Dr. Thompson moved his blog to the Unz Review. I was only reading the blog irregularly then so probably just never saw your comments on the IAB data. Mea culpa.

    About that move, it would be interesting to hear Dr. Thompson's thoughts (from a year and a half later) on the comments about moving to Unz in this: https://drjamesthompson.blogspot.com/2016/12/goodbye.html?m=1
    It is somewhat amusing that the first comment was about needing to moderate comments here and the last comment there was obvious spam.

    That was when I did the analysis. I’ve fuzzy memory about them now.
     
    Understood. I have the same problem. One thing I have taken to doing is putting my R analyses under source control (Git is very nicely integrated in RStudio) which makes it easier to go back and rerun/extend them and get a sense of the history. Being conscientious (or an*l ; ) about recording my data sources and transformations is also helpful. R Markdown provides a great mechanism for annotating, automating and recording data retrieval and transformation steps.

    Thank you for the Canada link and extended discussion. I agree that is a bit much for blog comments, but occasionally it is useful and I think your observations are on target.

    But what to do about that? Including short cycle tertiary education with University degrees doubles the percentage! It is important to consider that. Especially if trying to reconcile the OECD and IAB data.

    That was from a file most probably named “EAG_GRAD_ENTR_RATES_09022018023322785″ (unlikely to be the name I would type in).
     
    Did the contents of the file have any information about the original source? One weakness of CSVs (if that is what it was) is they tend not to have that sort of thing.

    not detailed calculations or citing every individual sources
     
    How about citing just one source? For the raw data. That seems a reasonable request. Then add any assumptions which you think have important effects on the results.

    Did you missed it? I said “I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve.”
     
    I saw that. I would ask if this is your profession, but that would be too personal. What you are describing is rather different from the aerospace engineer example you gave. I was thinking more of FP precision in calculations (e.g. numerical integration for finite element analyses of airframes). What you are describing is just data imprecision. Which is of course an issue. Just a different one.

    > I assume s and t are secondary and tertiary. What are n and z?

    For some forgotten reason I suffixed the first as ‘s’, thus the next one was naturally ‘t’. And the newer one ‘n’. ‘z’ is the last I would bother with from that particular source and year, it seems to be similar to the Table A1.1 with no decimal places.
     
    Thank you for elaborating, but do you realize that tells me nothing about the actual differences between those variables?

    I know you can request raw data but I prefer not to leave too many digital footprint.
     
    I understand (and share) that concern, but how is giving links to the OECD, IAB, etc. data an issue there?

    The simplifying assumptions are there to be able to arrived at some ball park figures or just simply throwing up your hands and gives up. The immigrants are already in the country and it is whether you want to know about their approximate abilities. Some just simply want to shut down the discussions.
     
    An excellent point. I think what I am trying to do is rather the opposite of trying to shut down discussion ; ). I totally agree with your approach, but for us to have a meaningful conversation about things like this it is critical for us to both be working from the same data and discussing the assumptions.

    You don’t seem to understand the difference between highlighting the RAW data used in the context of the discussion from the final results presentation. You seems to be pretty an*l.
     
    The line between an*l and conscientious can be a fine one. And I can certainly be guilty of stepping over it at times. However, I consider your ignoring factor of two or three effects in your analyses a much more important offense.

    I appreciate your analyses. They are creative and data based. However, without sources to the original data (to allow checking things like how the data was derived) and reasonable efforts to address relevant criticisms (which I consider mine above to be) they are much less useful than they might be otherwise.

    And for what it's worth, I consider a comment meant for others to read in this blog to be a place for something closer to final results presentation than raw data. What was the point of presenting 12 significant digits? I confess to being skeptical the original data contained that level of precision. I assume the OECD statisticians are not stupid. I thought you might have been making a joke at first (an oblique way of laughing at me being an*l), but you seem to have doubled down on it. Perhaps you did some calculations on the data and left the full precision in your results?

    P.S. Thank you for making a serious effort to look for your sources. I understand it can be a nuisance to try to reconstruct things like that.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  48. Ok boys and girls, I think we’ve flogged the higher math minutiae enough already. Let’s get back to the Mexico News Daily article and the group photo provided in same from my original inquiry back at #19:

    https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/?utm_source=Mexico+News+Daily&utm_campaign=197a5a9f33-newsletter_COPY_02&utm_medium=email&utm_term=0_f1536a3787-197a5a9f33-349530129

    We all should have a pretty good idea about what has been going on in the USA higher education rackets, credentialism and AA Diversity and Inclusion admissions/retention activities that are a significant component part of it all. We all should also be aware of the demographic changes going on here and in the white West in general.

    I don’t know whether or not the Mexico News Daily group photo accurately represents a cohort of applicants or newly admitted students (I do think it’s safe to say it doesn’t represent a group of rejects though) to the university discussed in the article or any other Mexican university for that matter.

    Now let’s acknowledge what has already been well covered regarding Mexican racial demographics
    (roughly White Euro 9%, Amerindian 27%, a wide ranging admixture Mestizo population of 62%,
    Black 1% and Arabic/Middle Eastern/Semite cluster of 1%) and Mexican socio-economic status by racial characteristics. One only need watch a few hours of Univision “news” shows and telenovelas to figure this out.

    So, my real curiosity lies into whether or not Mexico has adopted it’s own version of Social Justice Warfare and Diversity and Inclusion as it relates to university admissions, tuition and expenses financing and retention practices.

    Read More
    • Replies: @res

    So, my real curiosity lies into whether or not Mexico has adopted it’s own version of Social Justice Warfare and Diversity and Inclusion as it relates to university admissions, tuition and expenses financing and retention practices.
     
    An interesting question. This article is over 20 years old, but I think provides a clue: https://www.nytimes.com/1995/06/11/weekinreview/the-world-racism-mexico-s-in-denial.html

    But things are changing. In 2015 Mexico started tracking blacks: https://qz.com/569964/mexico-has-started-counting-its-afro-mexican-population/

    This 2014 article talks about affirmative action for women: https://www.wilsoncenter.org/article/mexico-seeks-to-empower-women-politics

    Perhaps try asking in Fred Reed's blog?

    P.S. It is generally a good thing to trim the tracking information out of links if possible. Here is your original link with that done: https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/

    For those who care about anonymity, those long, cryptic alphanumeric strings might be a unique identifier pointing at you.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  49. res says:
    @dux.ie
    > I actually learned about the Brain Drain data from Emil in a comment (a year ago today ; )

    I mentioned the IAB data in this blog back in 2016, comments 7,8,9. Dr. Thompson asked for the source.
    https://www.unz.com/jthompson/the-accuracy-of-stereotypes/
    That was when I did the analysis. I've fuzzy memory about them now.

    >I am not sure exactly which of those two categories the IAB includes in “highskilled.”

    IAB aggregated the census data from host countries. Canada is transparent about their immigrant selection process. In the Canadian process and I assume other countries' as well, 3-year diplomas are lumped together with university degrees,

    https://www.canada.ca/en/immigration-refugees-citizenship/services/immigrate-canada/express-entry/become-candidate/eligibility/education-assessed/read-report.html

    """University degree at the Doctoral (PhD) level 140 points
    University degree at the Master’s level or Licensed Professional degree 126 points"""

    """Bachelor's degree or other programs (three or more years) at a university, college, trade or technical school, or other institute 112 points
    ECA assessment results:
    College Diploma (three years)
    Diploma (three years)
    Three-year diploma in [name of discipline]
    Three-year advanced diploma
    Bachelor's Degree ... """

    Even the two years diplomas or extended high school diplomas are not that far behind, still categorized them as two-year degree, this could be the “Short-cycle tertiary”

    """Two-year degree, diploma or certificate from a program at a university, college, trade or technical school, or other institute 91 points
    ECA assessment results:
    College Diploma
    College Diploma (two years)
    Diploma (two years)
    Two-year diploma
    Two-year diploma in [name of discipline]
    Secondary school diploma and diploma (two years)
    Associate Degree
    Associate of [Arts/Science] degree
    Post-secondary Diploma with a focus in [area of concentration]
    Two-year post-secondary Diploma with a focus in [area of concentration]
    Two-year associate degree"""

    This could be the “Post-secondary non-tertiary”

    """One-year degree, diploma or certificate from a program at a university, college, trade or technical school, or other institute 84 points
    ECA assessment results:
    Certificate ... """

    > It would be great if you could provide a reference

    That was from a file most probably named "EAG_GRAD_ENTR_RATES_09022018023322785" (unlikely to be the name I would type in). Google search comes up empty. The data has values for W1 which they termed "Mobile Students" and I assumed that to be oversea students, and W0 is "World (all entities, including reference area, including IO)" I assumed to be the total number of graduates from that country dispite the word "World". Google seach comes up with some hits for that term. None of them from OECD. Seems to be a standard EU category. The values of W0-W1 listed is similar to that of table A1.1. That was quite some time ago, my memory is a bit fuzzy about that. I only saved a csv copy. I have a few thumb drives bit rot in the last 2 years. Beware of them as you will only realize that when you open them up and they are full of nulls while the file sizes still remain the same. Backing up of null files are pretty useless.

    > I assume s and t are secondary and tertiary. What are n and z?

    For some forgotten reason I suffixed the first as 's', thus the next one was naturally 't'. And the newer one 'n'. 'z' is the last I would bother with from that particular source and year, it seems to be similar to the Table A1.1 with no decimal places.

    > d and x

    Missing values. Cannot filled in from external sources. These UniA values are from UNESCO or country census,

    ISO|Country|UniA
    BGD|Bangladesh|2.3
    HKG|HongKong|17.21
    IND|India|5.64
    IRN|Iran|18.53
    JAM|Jamaica|8.4
    NGA|Nigeria|7.

    Look. Comment section can only talk about overall methodology and that was already too long, not detailed calculations or citing every individual sources. Even on methodology I already said too much about the initial assumptions. Elsewhere the discussions had been bogged down on the assumption of frictionless uni student selections. The simplifying assumptions are there to be able to arrived at some ball park figures or just simply throwing up your hands and gives up. The immigrants are already in the country and it is whether you want to know about their approximate abilities. Some just simply want to shut down the discussions.

    > Have you ever had to care about floating point precision professionally? (I have)

    Did you missed it? I said "I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve." For example, many of Pew research data with small percentages less than 5 had no decimal places which rendered them useless for further exploration. I know you can request raw data but I prefer not to leave too many digital footprint.

    > P.S. One clue that you are not an engineer-quoting 12 significant digits for numbers that would be lucky to be accurate to 3. If you are depending on that precision you are a fool.

    You don't seem to understand the difference between highlighting the RAW data used in the context of the discussion from the final results presentation. You seems to be pretty an*l.

    I mentioned the IAB data in this blog back in 2016, comments 7,8,9. Dr. Thompson asked for the source.

    Sorry that I missed that. Sometimes learning about something requires having a need for it, and the IAB data was very useful in Chanda Chisala threads that happened in the interim.

    On reflection, I notice that that post was one of the last before Dr. Thompson moved his blog to the Unz Review. I was only reading the blog irregularly then so probably just never saw your comments on the IAB data. Mea culpa.

    About that move, it would be interesting to hear Dr. Thompson’s thoughts (from a year and a half later) on the comments about moving to Unz in this: https://drjamesthompson.blogspot.com/2016/12/goodbye.html?m=1
    It is somewhat amusing that the first comment was about needing to moderate comments here and the last comment there was obvious spam.

    That was when I did the analysis. I’ve fuzzy memory about them now.

    Understood. I have the same problem. One thing I have taken to doing is putting my R analyses under source control (Git is very nicely integrated in RStudio) which makes it easier to go back and rerun/extend them and get a sense of the history. Being conscientious (or an*l ; ) about recording my data sources and transformations is also helpful. R Markdown provides a great mechanism for annotating, automating and recording data retrieval and transformation steps.

    Thank you for the Canada link and extended discussion. I agree that is a bit much for blog comments, but occasionally it is useful and I think your observations are on target.

    But what to do about that? Including short cycle tertiary education with University degrees doubles the percentage! It is important to consider that. Especially if trying to reconcile the OECD and IAB data.

    That was from a file most probably named “EAG_GRAD_ENTR_RATES_09022018023322785″ (unlikely to be the name I would type in).

    Did the contents of the file have any information about the original source? One weakness of CSVs (if that is what it was) is they tend not to have that sort of thing.

    not detailed calculations or citing every individual sources

    How about citing just one source? For the raw data. That seems a reasonable request. Then add any assumptions which you think have important effects on the results.

    Did you missed it? I said “I am getting too many 0 values when using double precision to do calculations around the right hand tail of the bell curve.”

    I saw that. I would ask if this is your profession, but that would be too personal. What you are describing is rather different from the aerospace engineer example you gave. I was thinking more of FP precision in calculations (e.g. numerical integration for finite element analyses of airframes). What you are describing is just data imprecision. Which is of course an issue. Just a different one.

    > I assume s and t are secondary and tertiary. What are n and z?

    For some forgotten reason I suffixed the first as ‘s’, thus the next one was naturally ‘t’. And the newer one ‘n’. ‘z’ is the last I would bother with from that particular source and year, it seems to be similar to the Table A1.1 with no decimal places.

    Thank you for elaborating, but do you realize that tells me nothing about the actual differences between those variables?

    I know you can request raw data but I prefer not to leave too many digital footprint.

    I understand (and share) that concern, but how is giving links to the OECD, IAB, etc. data an issue there?

    The simplifying assumptions are there to be able to arrived at some ball park figures or just simply throwing up your hands and gives up. The immigrants are already in the country and it is whether you want to know about their approximate abilities. Some just simply want to shut down the discussions.

    An excellent point. I think what I am trying to do is rather the opposite of trying to shut down discussion ; ). I totally agree with your approach, but for us to have a meaningful conversation about things like this it is critical for us to both be working from the same data and discussing the assumptions.

    You don’t seem to understand the difference between highlighting the RAW data used in the context of the discussion from the final results presentation. You seems to be pretty an*l.

    The line between an*l and conscientious can be a fine one. And I can certainly be guilty of stepping over it at times. However, I consider your ignoring factor of two or three effects in your analyses a much more important offense.

    I appreciate your analyses. They are creative and data based. However, without sources to the original data (to allow checking things like how the data was derived) and reasonable efforts to address relevant criticisms (which I consider mine above to be) they are much less useful than they might be otherwise.

    And for what it’s worth, I consider a comment meant for others to read in this blog to be a place for something closer to final results presentation than raw data. What was the point of presenting 12 significant digits? I confess to being skeptical the original data contained that level of precision. I assume the OECD statisticians are not stupid. I thought you might have been making a joke at first (an oblique way of laughing at me being an*l), but you seem to have doubled down on it. Perhaps you did some calculations on the data and left the full precision in your results?

    P.S. Thank you for making a serious effort to look for your sources. I understand it can be a nuisance to try to reconstruct things like that.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  50. res says:
    @Jim Bob Lassiter
    Ok boys and girls, I think we've flogged the higher math minutiae enough already. Let's get back to the Mexico News Daily article and the group photo provided in same from my original inquiry back at #19:

    https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/?utm_source=Mexico+News+Daily&utm_campaign=197a5a9f33-newsletter_COPY_02&utm_medium=email&utm_term=0_f1536a3787-197a5a9f33-349530129


    We all should have a pretty good idea about what has been going on in the USA higher education rackets, credentialism and AA Diversity and Inclusion admissions/retention activities that are a significant component part of it all. We all should also be aware of the demographic changes going on here and in the white West in general.

    I don't know whether or not the Mexico News Daily group photo accurately represents a cohort of applicants or newly admitted students (I do think it's safe to say it doesn't represent a group of rejects though) to the university discussed in the article or any other Mexican university for that matter.

    Now let's acknowledge what has already been well covered regarding Mexican racial demographics
    (roughly White Euro 9%, Amerindian 27%, a wide ranging admixture Mestizo population of 62%,
    Black 1% and Arabic/Middle Eastern/Semite cluster of 1%) and Mexican socio-economic status by racial characteristics. One only need watch a few hours of Univision "news" shows and telenovelas to figure this out.

    So, my real curiosity lies into whether or not Mexico has adopted it's own version of Social Justice Warfare and Diversity and Inclusion as it relates to university admissions, tuition and expenses financing and retention practices.

    So, my real curiosity lies into whether or not Mexico has adopted it’s own version of Social Justice Warfare and Diversity and Inclusion as it relates to university admissions, tuition and expenses financing and retention practices.

    An interesting question. This article is over 20 years old, but I think provides a clue: https://www.nytimes.com/1995/06/11/weekinreview/the-world-racism-mexico-s-in-denial.html

    But things are changing. In 2015 Mexico started tracking blacks: https://qz.com/569964/mexico-has-started-counting-its-afro-mexican-population/

    This 2014 article talks about affirmative action for women: https://www.wilsoncenter.org/article/mexico-seeks-to-empower-women-politics

    Perhaps try asking in Fred Reed’s blog?

    P.S. It is generally a good thing to trim the tracking information out of links if possible. Here is your original link with that done: https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/

    For those who care about anonymity, those long, cryptic alphanumeric strings might be a unique identifier pointing at you.

    Read More
    • Replies: @Jim Bob Lassiter
    So, do you suppose an honest follow up article referring to the 20 year NYT Depalma article would ever see the light of Day at the Carlos Slim Times?
    , @Jim Bob Lassiter
    "Perhaps try asking in Fred Reed’s blog?"

    Interesting proposition seeing that he has a new piece up about dumbass Gringos. I think I'll let sleeping dogs lay on that one though. We've all heard about his little Wakandan Gringo Shangri-La on Lake Chapala with all of its outliers. I'm sure Fred didn't form his new family with the first maids to make up his bed either. Let's give credit where credit is due.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  51. @res

    So, my real curiosity lies into whether or not Mexico has adopted it’s own version of Social Justice Warfare and Diversity and Inclusion as it relates to university admissions, tuition and expenses financing and retention practices.
     
    An interesting question. This article is over 20 years old, but I think provides a clue: https://www.nytimes.com/1995/06/11/weekinreview/the-world-racism-mexico-s-in-denial.html

    But things are changing. In 2015 Mexico started tracking blacks: https://qz.com/569964/mexico-has-started-counting-its-afro-mexican-population/

    This 2014 article talks about affirmative action for women: https://www.wilsoncenter.org/article/mexico-seeks-to-empower-women-politics

    Perhaps try asking in Fred Reed's blog?

    P.S. It is generally a good thing to trim the tracking information out of links if possible. Here is your original link with that done: https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/

    For those who care about anonymity, those long, cryptic alphanumeric strings might be a unique identifier pointing at you.

    So, do you suppose an honest follow up article referring to the 20 year NYT Depalma article would ever see the light of Day at the Carlos Slim Times?

    Read More
    • Replies: @res
    Heh. I'm pretty sure your question was rhetorical, but to play along. No. Unless they could find a way to spin it to make Trump look bad somehow.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  52. @res

    So, my real curiosity lies into whether or not Mexico has adopted it’s own version of Social Justice Warfare and Diversity and Inclusion as it relates to university admissions, tuition and expenses financing and retention practices.
     
    An interesting question. This article is over 20 years old, but I think provides a clue: https://www.nytimes.com/1995/06/11/weekinreview/the-world-racism-mexico-s-in-denial.html

    But things are changing. In 2015 Mexico started tracking blacks: https://qz.com/569964/mexico-has-started-counting-its-afro-mexican-population/

    This 2014 article talks about affirmative action for women: https://www.wilsoncenter.org/article/mexico-seeks-to-empower-women-politics

    Perhaps try asking in Fred Reed's blog?

    P.S. It is generally a good thing to trim the tracking information out of links if possible. Here is your original link with that done: https://mexiconewsdaily.com/news/92000-wrote-university-entrance-exam-but-only-room-for-24000/

    For those who care about anonymity, those long, cryptic alphanumeric strings might be a unique identifier pointing at you.

    “Perhaps try asking in Fred Reed’s blog?”

    Interesting proposition seeing that he has a new piece up about dumbass Gringos. I think I’ll let sleeping dogs lay on that one though. We’ve all heard about his little Wakandan Gringo Shangri-La on Lake Chapala with all of its outliers. I’m sure Fred didn’t form his new family with the first maids to make up his bed either. Let’s give credit where credit is due.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  53. res says:
    @Jim Bob Lassiter
    So, do you suppose an honest follow up article referring to the 20 year NYT Depalma article would ever see the light of Day at the Carlos Slim Times?

    Heh. I’m pretty sure your question was rhetorical, but to play along. No. Unless they could find a way to spin it to make Trump look bad somehow.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  54. dux.ie says:
    @James Thompson
    Interesting work @dux.ie and @res, and thanks for telling us about it. Couple of problems. India is very variable in ability. Provinces vary enormously, and also by castes and clans, the latter in ways which may be hard to track down without Indian specialists to explain them. PISA helps, but countries tend to cheat a bit. Mingrui Wang has data which suggests estimates for Chinese intelligence may be inflated by using almost exclusively urban samples. Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai (this from memory, but certainly a small sample which was university based).
    So, I like the method of limits, but countries differ in how honestly and consistently they grant university entrance. None of this is fatal to estimation, but it all causes some anomalies.
    Also, UK immigration can be weird. Absolute jumble of abilities, as you would expect if they were not selecting for ability.

    > Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai

    There are always people whinging about the PISA 2012 Shanghai results being from a urban area and needed to scale down for China. The Piffer’s results of the alleged Shanghai University graduates’ genomes under-predicting the Chinese intelligence might be closer to the truth.

    First, it is not always true that city students perform better than the rural students. The American ivy league professor without any proof or data raised that question when the US NAEP test results showed the rural was doing better than city, town and suburban, table 6b. https://nces.ed.gov/programs/coe/pdf/coe_tla.pdf Are there any Chinese data? Recently the Chinese Education Dept issued an directive against any sources highlighting the high performance students in the College Entrance Exams (CEE GaoKao). So there will be no new data and the existing data will be progressively removed. Thus try you luck on this photo,

    the top two students were the top scorers for the CEE science (693) and arts (629) from the counties while the bottom two were that from the cities (respectively 685, 613). Both the counties students beat the city students in their respective majors. The four mentioned students were the top students from rural counties and cities among the about 10 million students sat the CEE. By sheer number out of 10 million students they can afford to select only 0.03% (IQlike=178) to be admitted to the top 2 Peking and Tsinghua Universities.

    Are there any more global data beside anecdotal evidence? Dr. Hongbin Li, James Liang Director of the China Program, Stanford Center on Global Poverty and Development, Stanford University, showed the statistics of the Chinese CEE for all students in the urban and rural areas, https://www.youtube.com/watch?v=lP3QW8OgVJE The data at 16:45 the screen capture,

    http://tinypic.com/view.php?pic=2ilhi83&s=9

    The results for the urban students seem to agree to the assumption that most of them are descendents of the ancient Chinese imperial court officials who passed the imperial court exams writing the “eight legged essays”. They greatly excelled in languages like Mandarin and English whereas most of those from the rural areas greatly excelled in mathematics. There is no centralized CEE and each administrative regions set their own exams and the level of the maximum scores. Thus the national average is the average of the provincial average with respect to their maximum scores. However I still do not understand why the sum of the rural and city averages come out to 1. Anyway, for the total scores and the composite scores the rural students performed better than the urban students on a national scale. Thus as Piffer found out, if the Chinese PISA score is to be modified it might have to move upwards.

    As to the quality of the Shanghai university students, that depends on the CEE scores distribution and the number of available local universities. The governance of the Chinese administrative regions is pretty autonomous in certain areas like education. Thus rich regions can finance more universities which have preference for local students with the result that the minimum entrance scores will be reduced. Shanghai being a rich region no doubt can attract many top students. However, with many universities there, the resulting average student quality is not that good. From a ranked list of the cut-off percentage off the respective max provincial scores ratio1, Shanghai ranked third from the bottom, or weighted by the simple ratio of IQ over the national average IQ, W1, rank 16 just below the middle.

    Rank|W1|ratio1|RegionE
    1|86.2|76.9|Zhejiang
    2|81.7|74.0|Beijing
    3|73.7|70.9|Liaoning
    16|67.9|60.9|Shanghai

    Read More
    • Replies: @James Thompson
    Thanks. I just got all the province scores for IQ test scores aged 8-10 which is highly relevant for the provincial estimates and their contribution to the national totals. Hope to get this into a good form later.
    , @res
    Thanks for the interesting and well referenced comment. Here is some background on the CEE for those (like me) unfamiliar with it: https://en.wikipedia.org/wiki/National_Higher_Education_Entrance_Examination

    One thing that struck me about your photo of the CEE high scorers is how much ceiling the test appears to have. If I understand correctly, the test has a maximum score of 750 yet the highest score we see above is 693!

    Am I correct in inferring that the decimal results in your screenshot are percentiles of the maximum? Are there any mean/SD statistics for the CEE?

    I was surprised by this at the Wiki link above:

    In 2006, a record high of 9.5 million people applied for tertiary education entry in China. Of these, 8.8 million (93%) took the national entrance exam and 27,600 (0.28%) were exempted (保送) due to exceptional or special talent. Everyone else (700,000 students) took other standardized entrance exams, such as those designed for adult education students.
     
    Why would they exempt special talents from taking the test rather than just giving them special preference beyond the test scores? Exempting them makes me wonder if this is a mechanism for sneaking in grossly unqualified people with connections.

    By sheer number out of 10 million students they can afford to select only 0.03% (IQlike=178)
     
    I was a bit confused by this. If I understand correctly, 0.03% is roughly 4 SD which would be a 160 deviation IQ: https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule#Table_of_numerical_values
    Can you elaborate on how you got the 178 IQlike number? (by my math 5 SD = 175 IQ is roughly 2 orders of magnitude rarer)

    Are you correcting for the test taking population having an average IQ higher than 100? That could explain the result, but raises questions about how that norming is done.

    P.S. Any idea what proportion of the age cohort takes the CEE? If you are used to China as a baseline that might explain an assumption that most students take the admissions tests (which I don't think is true of Mexico).
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  55. @dux.ie
    > Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai

    There are always people whinging about the PISA 2012 Shanghai results being from a urban area and needed to scale down for China. The Piffer's results of the alleged Shanghai University graduates' genomes under-predicting the Chinese intelligence might be closer to the truth.

    First, it is not always true that city students perform better than the rural students. The American ivy league professor without any proof or data raised that question when the US NAEP test results showed the rural was doing better than city, town and suburban, table 6b. https://nces.ed.gov/programs/coe/pdf/coe_tla.pdf Are there any Chinese data? Recently the Chinese Education Dept issued an directive against any sources highlighting the high performance students in the College Entrance Exams (CEE GaoKao). So there will be no new data and the existing data will be progressively removed. Thus try you luck on this photo,
    https://www.whatsonweibo.com/wp-content/uploads/2018/05/917aaba497edfe6e-422x360.jpg
    the top two students were the top scorers for the CEE science (693) and arts (629) from the counties while the bottom two were that from the cities (respectively 685, 613). Both the counties students beat the city students in their respective majors. The four mentioned students were the top students from rural counties and cities among the about 10 million students sat the CEE. By sheer number out of 10 million students they can afford to select only 0.03% (IQlike=178) to be admitted to the top 2 Peking and Tsinghua Universities.

    Are there any more global data beside anecdotal evidence? Dr. Hongbin Li, James Liang Director of the China Program, Stanford Center on Global Poverty and Development, Stanford University, showed the statistics of the Chinese CEE for all students in the urban and rural areas, https://www.youtube.com/watch?v=lP3QW8OgVJE The data at 16:45 the screen capture,
    http://tinypic.com/view.php?pic=2ilhi83&s=9
    The results for the urban students seem to agree to the assumption that most of them are descendents of the ancient Chinese imperial court officials who passed the imperial court exams writing the "eight legged essays". They greatly excelled in languages like Mandarin and English whereas most of those from the rural areas greatly excelled in mathematics. There is no centralized CEE and each administrative regions set their own exams and the level of the maximum scores. Thus the national average is the average of the provincial average with respect to their maximum scores. However I still do not understand why the sum of the rural and city averages come out to 1. Anyway, for the total scores and the composite scores the rural students performed better than the urban students on a national scale. Thus as Piffer found out, if the Chinese PISA score is to be modified it might have to move upwards.

    As to the quality of the Shanghai university students, that depends on the CEE scores distribution and the number of available local universities. The governance of the Chinese administrative regions is pretty autonomous in certain areas like education. Thus rich regions can finance more universities which have preference for local students with the result that the minimum entrance scores will be reduced. Shanghai being a rich region no doubt can attract many top students. However, with many universities there, the resulting average student quality is not that good. From a ranked list of the cut-off percentage off the respective max provincial scores ratio1, Shanghai ranked third from the bottom, or weighted by the simple ratio of IQ over the national average IQ, W1, rank 16 just below the middle.

    Rank|W1|ratio1|RegionE
    1|86.2|76.9|Zhejiang
    2|81.7|74.0|Beijing
    3|73.7|70.9|Liaoning
    16|67.9|60.9|Shanghai

    Thanks. I just got all the province scores for IQ test scores aged 8-10 which is highly relevant for the provincial estimates and their contribution to the national totals. Hope to get this into a good form later.

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

    Putting aside artists of one sort or another, and pure mathematicians, you could argue that the greatest minds of the last few hundred years have been Isaac Newton, Albert Einstein and James Clerk Maxwell. I assume Einstein was urban; Newton and Maxwell were rural.

    Newton was also a pure mathematician of the highest class even if not numero uno.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  57. res says:
    @dux.ie
    > Davide Piffer found that his equation under-predicted Chinese intelligence, and then found that the Chinese sample in the 1000 Genomes database seemed to come from university graduates in Shanghai

    There are always people whinging about the PISA 2012 Shanghai results being from a urban area and needed to scale down for China. The Piffer's results of the alleged Shanghai University graduates' genomes under-predicting the Chinese intelligence might be closer to the truth.

    First, it is not always true that city students perform better than the rural students. The American ivy league professor without any proof or data raised that question when the US NAEP test results showed the rural was doing better than city, town and suburban, table 6b. https://nces.ed.gov/programs/coe/pdf/coe_tla.pdf Are there any Chinese data? Recently the Chinese Education Dept issued an directive against any sources highlighting the high performance students in the College Entrance Exams (CEE GaoKao). So there will be no new data and the existing data will be progressively removed. Thus try you luck on this photo,
    https://www.whatsonweibo.com/wp-content/uploads/2018/05/917aaba497edfe6e-422x360.jpg
    the top two students were the top scorers for the CEE science (693) and arts (629) from the counties while the bottom two were that from the cities (respectively 685, 613). Both the counties students beat the city students in their respective majors. The four mentioned students were the top students from rural counties and cities among the about 10 million students sat the CEE. By sheer number out of 10 million students they can afford to select only 0.03% (IQlike=178) to be admitted to the top 2 Peking and Tsinghua Universities.

    Are there any more global data beside anecdotal evidence? Dr. Hongbin Li, James Liang Director of the China Program, Stanford Center on Global Poverty and Development, Stanford University, showed the statistics of the Chinese CEE for all students in the urban and rural areas, https://www.youtube.com/watch?v=lP3QW8OgVJE The data at 16:45 the screen capture,
    http://tinypic.com/view.php?pic=2ilhi83&s=9
    The results for the urban students seem to agree to the assumption that most of them are descendents of the ancient Chinese imperial court officials who passed the imperial court exams writing the "eight legged essays". They greatly excelled in languages like Mandarin and English whereas most of those from the rural areas greatly excelled in mathematics. There is no centralized CEE and each administrative regions set their own exams and the level of the maximum scores. Thus the national average is the average of the provincial average with respect to their maximum scores. However I still do not understand why the sum of the rural and city averages come out to 1. Anyway, for the total scores and the composite scores the rural students performed better than the urban students on a national scale. Thus as Piffer found out, if the Chinese PISA score is to be modified it might have to move upwards.

    As to the quality of the Shanghai university students, that depends on the CEE scores distribution and the number of available local universities. The governance of the Chinese administrative regions is pretty autonomous in certain areas like education. Thus rich regions can finance more universities which have preference for local students with the result that the minimum entrance scores will be reduced. Shanghai being a rich region no doubt can attract many top students. However, with many universities there, the resulting average student quality is not that good. From a ranked list of the cut-off percentage off the respective max provincial scores ratio1, Shanghai ranked third from the bottom, or weighted by the simple ratio of IQ over the national average IQ, W1, rank 16 just below the middle.

    Rank|W1|ratio1|RegionE
    1|86.2|76.9|Zhejiang
    2|81.7|74.0|Beijing
    3|73.7|70.9|Liaoning
    16|67.9|60.9|Shanghai

    Thanks for the interesting and well referenced comment. Here is some background on the CEE for those (like me) unfamiliar with it: https://en.wikipedia.org/wiki/National_Higher_Education_Entrance_Examination

    One thing that struck me about your photo of the CEE high scorers is how much ceiling the test appears to have. If I understand correctly, the test has a maximum score of 750 yet the highest score we see above is 693!

    Am I correct in inferring that the decimal results in your screenshot are percentiles of the maximum? Are there any mean/SD statistics for the CEE?

    I was surprised by this at the Wiki link above:

    In 2006, a record high of 9.5 million people applied for tertiary education entry in China. Of these, 8.8 million (93%) took the national entrance exam and 27,600 (0.28%) were exempted (保送) due to exceptional or special talent. Everyone else (700,000 students) took other standardized entrance exams, such as those designed for adult education students.

    Why would they exempt special talents from taking the test rather than just giving them special preference beyond the test scores? Exempting them makes me wonder if this is a mechanism for sneaking in grossly unqualified people with connections.

    By sheer number out of 10 million students they can afford to select only 0.03% (IQlike=178)

    I was a bit confused by this. If I understand correctly, 0.03% is roughly 4 SD which would be a 160 deviation IQ: https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule#Table_of_numerical_values
    Can you elaborate on how you got the 178 IQlike number? (by my math 5 SD = 175 IQ is roughly 2 orders of magnitude rarer)

    Are you correcting for the test taking population having an average IQ higher than 100? That could explain the result, but raises questions about how that norming is done.

    P.S. Any idea what proportion of the age cohort takes the CEE? If you are used to China as a baseline that might explain an assumption that most students take the admissions tests (which I don’t think is true of Mexico).

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  58. dux.ie says:

    > Can you elaborate on how you got the 178 IQlike number?

    Caffeine deficiency. Thought of correcting it today. You beat me to it. Its about IQlike=158.

    > If I understand correctly, the test has a maximum score of 750 yet the highest score we see above is 693!

    I got the impression that they really want to find the very top students. The CEE score distribution curve from Sichuan (got the jpg but cannot find the url again) might be the typical where by inspection only a handful got above 650 with plenty of room to the ceiling of 750. The left hand tail is relatively long allowing finer grain cut-off points for various other applications. The uni student selection is on the right hand tail. It is similar to the GRE score distribution. The ACT score distribution is the reverse, longer right hand tail and squashed left hand tail, might be the result of “no child left behind” and that might be why some regions prefer ACT. Squashed tails reduce ethnic gaps. The SAT distribution is trying to be bell shape.

    The Australian Monash Uni directly accept GaoKao results and the provincial cut-off and max scores are listed, https://www.monash.edu/__data/assets/pdf_file/0005/57668/gaokao-flyer-for-students.pdf The loop hole being it is easier for someone from Shanghai with cut-off ratio third from bottom at 60.9% (29 out 31) than those from Beijing at 2nd from top at 74.0%

    For cross checking the PISA results the Middle School Exam (ZhongKoa) for grade 9 is the closest to the 15 yo mostly in grade 10. I suspect the ZhongKoa caused many to repeat grade 9. The PISA 15 yo in G10 is 54.19% and that for G09 is 39.56% For G12 it is 0.08% and may be half of these will have express entry to Peking and Tsinghua universities. In Massachusetts G12 is 0.07% but for USA it is .22%

    > Are there any mean/SD statistics for the CEE?
    I suspect that Stanford researcher is a policy wonk not that good on Stats. The numbers he shown could be the respective proportions that met the cut-off point and so they added up to 1. The ratio of rural to city populations is 590/793. Since the performance values for the rural students are higher, so I accepted his words at face value that the rural performed better than that for the city. No SD values.

    There is too much emphasis on the difference between rural and city students. Poor rural areas usually do not have senior high schools and the migrant children usually follow their parents to live in the cities and enter special senior high schools for migrant children in the cities but they are still being classified as from rural areas.

    > Of these, 8.8 million (93%) took the national entrance exam and 27,600 (0.28%) were exempted (保送) due to exceptional or special talent. Why would they exempt special talents from taking the test rather than just giving them special preference beyond the test scores?

    They are pretty transparent about those exemptions. I was trying to get a sample of the students admitted to get the surname distribution before I found out that they were actually lists of those exceptions. There are only two type of exemptions. One is ex-Olympic competitors and they are exclusively sent to Peking and Tsinghua Universities mostly in those elite streams like STEM and Medicine and were given grades of excellent. The others are language specialist trainees for the government or big corporations. Even those sent to the Aeronautical University are language trainees specialized in those areas. They seems to be very serious about specialization. Olympic or any other world competitors will be training full time. Like the World Women Chess champion Hou had said that she was advised not to undertake university study. She ignored that and she admitted that her chess performance was affected.

    Well that sample of language trainees gave me some ideas on the relative importance of some countries relative to China.

    Rank|N|Pct|Language
    1|287|36.5|English
    2|101|12.8|German
    3|91|11.6|Japanese
    4|78|9.9|French
    5|57|7.2|Spanish
    9|19|2.4|Portuguese # Mexico and Caribbean more important than Brazil??
    17|3|0.4|Latin # There are big negotiations with Vatican that upset HK Cardinal Zen.
    24|1|0.1|Tamil # I suppose Indian diplomats speak English
    25|1|0.1|Sanskrit

    The chairman of the Asian Infrastructure Investment Bank AIIB could be formerly one of those trainees. He is still an adjunt professor in those language schools.

    Read More
    • Replies: @Jim Bob Lassiter
    Where does Russian place in Chinese foreign language proficiency priorities?
    , @res
    Thank you for all of the additional information! The Monash PDF has an interesting Table 1 showing how much admissions requirements vary by province. For interpreting the following data I think it is worthwhile to know that Monash has significantly lower than average thresholds for admission for Fujian province (489 for full admission). I don't understand why Shanghai province has the lowest thresholds (except for ethnic Tibetans) with 402 for full admission.

    This paper about gender gaps in the Gaokao has statistical data for 8164 people from Anxi county in Fujian province: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2717464
    Figure 1 has distribution plots and Table 1 has summary statistics. Mean (SD) for the Gaokao in Anxi were 426.34 (80.69) so in theory the ceiling is ~4 SD for that group.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  59. @dux.ie
    > Can you elaborate on how you got the 178 IQlike number?

    Caffeine deficiency. Thought of correcting it today. You beat me to it. Its about IQlike=158.

    > If I understand correctly, the test has a maximum score of 750 yet the highest score we see above is 693!

    I got the impression that they really want to find the very top students. The CEE score distribution curve from Sichuan (got the jpg but cannot find the url again) might be the typical where by inspection only a handful got above 650 with plenty of room to the ceiling of 750. The left hand tail is relatively long allowing finer grain cut-off points for various other applications. The uni student selection is on the right hand tail. It is similar to the GRE score distribution. The ACT score distribution is the reverse, longer right hand tail and squashed left hand tail, might be the result of "no child left behind" and that might be why some regions prefer ACT. Squashed tails reduce ethnic gaps. The SAT distribution is trying to be bell shape.

    The Australian Monash Uni directly accept GaoKao results and the provincial cut-off and max scores are listed, https://www.monash.edu/__data/assets/pdf_file/0005/57668/gaokao-flyer-for-students.pdf The loop hole being it is easier for someone from Shanghai with cut-off ratio third from bottom at 60.9% (29 out 31) than those from Beijing at 2nd from top at 74.0%

    For cross checking the PISA results the Middle School Exam (ZhongKoa) for grade 9 is the closest to the 15 yo mostly in grade 10. I suspect the ZhongKoa caused many to repeat grade 9. The PISA 15 yo in G10 is 54.19% and that for G09 is 39.56% For G12 it is 0.08% and may be half of these will have express entry to Peking and Tsinghua universities. In Massachusetts G12 is 0.07% but for USA it is .22%

    > Are there any mean/SD statistics for the CEE?
    I suspect that Stanford researcher is a policy wonk not that good on Stats. The numbers he shown could be the respective proportions that met the cut-off point and so they added up to 1. The ratio of rural to city populations is 590/793. Since the performance values for the rural students are higher, so I accepted his words at face value that the rural performed better than that for the city. No SD values.

    There is too much emphasis on the difference between rural and city students. Poor rural areas usually do not have senior high schools and the migrant children usually follow their parents to live in the cities and enter special senior high schools for migrant children in the cities but they are still being classified as from rural areas.

    > Of these, 8.8 million (93%) took the national entrance exam and 27,600 (0.28%) were exempted (保送) due to exceptional or special talent. Why would they exempt special talents from taking the test rather than just giving them special preference beyond the test scores?

    They are pretty transparent about those exemptions. I was trying to get a sample of the students admitted to get the surname distribution before I found out that they were actually lists of those exceptions. There are only two type of exemptions. One is ex-Olympic competitors and they are exclusively sent to Peking and Tsinghua Universities mostly in those elite streams like STEM and Medicine and were given grades of excellent. The others are language specialist trainees for the government or big corporations. Even those sent to the Aeronautical University are language trainees specialized in those areas. They seems to be very serious about specialization. Olympic or any other world competitors will be training full time. Like the World Women Chess champion Hou had said that she was advised not to undertake university study. She ignored that and she admitted that her chess performance was affected.

    Well that sample of language trainees gave me some ideas on the relative importance of some countries relative to China.

    Rank|N|Pct|Language
    1|287|36.5|English
    2|101|12.8|German
    3|91|11.6|Japanese
    4|78|9.9|French
    5|57|7.2|Spanish
    9|19|2.4|Portuguese # Mexico and Caribbean more important than Brazil??
    17|3|0.4|Latin # There are big negotiations with Vatican that upset HK Cardinal Zen.
    24|1|0.1|Tamil # I suppose Indian diplomats speak English
    25|1|0.1|Sanskrit

    The chairman of the Asian Infrastructure Investment Bank AIIB could be formerly one of those trainees. He is still an adjunt professor in those language schools.

    Where does Russian place in Chinese foreign language proficiency priorities?

    Read More
    • Replies: @dux.ie
    Rank|N|Pct|Language
    7|48|6.1|Russian
    23|1|0.1|Zulu

    Another interesting data is that Zulu is the only African language prioritized. China gives lots of scholarships to the Africans and I suppose the official communications will be in French, English or Mandarin. I used to know a Senegalese graduate from Peking Uni computer science major and our party trick was to let him ordered Chinese meals for us from the Cantonese speaking waiters who could not understand Mandarin.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  60. res says:
    @dux.ie
    > Can you elaborate on how you got the 178 IQlike number?

    Caffeine deficiency. Thought of correcting it today. You beat me to it. Its about IQlike=158.

    > If I understand correctly, the test has a maximum score of 750 yet the highest score we see above is 693!

    I got the impression that they really want to find the very top students. The CEE score distribution curve from Sichuan (got the jpg but cannot find the url again) might be the typical where by inspection only a handful got above 650 with plenty of room to the ceiling of 750. The left hand tail is relatively long allowing finer grain cut-off points for various other applications. The uni student selection is on the right hand tail. It is similar to the GRE score distribution. The ACT score distribution is the reverse, longer right hand tail and squashed left hand tail, might be the result of "no child left behind" and that might be why some regions prefer ACT. Squashed tails reduce ethnic gaps. The SAT distribution is trying to be bell shape.

    The Australian Monash Uni directly accept GaoKao results and the provincial cut-off and max scores are listed, https://www.monash.edu/__data/assets/pdf_file/0005/57668/gaokao-flyer-for-students.pdf The loop hole being it is easier for someone from Shanghai with cut-off ratio third from bottom at 60.9% (29 out 31) than those from Beijing at 2nd from top at 74.0%

    For cross checking the PISA results the Middle School Exam (ZhongKoa) for grade 9 is the closest to the 15 yo mostly in grade 10. I suspect the ZhongKoa caused many to repeat grade 9. The PISA 15 yo in G10 is 54.19% and that for G09 is 39.56% For G12 it is 0.08% and may be half of these will have express entry to Peking and Tsinghua universities. In Massachusetts G12 is 0.07% but for USA it is .22%

    > Are there any mean/SD statistics for the CEE?
    I suspect that Stanford researcher is a policy wonk not that good on Stats. The numbers he shown could be the respective proportions that met the cut-off point and so they added up to 1. The ratio of rural to city populations is 590/793. Since the performance values for the rural students are higher, so I accepted his words at face value that the rural performed better than that for the city. No SD values.

    There is too much emphasis on the difference between rural and city students. Poor rural areas usually do not have senior high schools and the migrant children usually follow their parents to live in the cities and enter special senior high schools for migrant children in the cities but they are still being classified as from rural areas.

    > Of these, 8.8 million (93%) took the national entrance exam and 27,600 (0.28%) were exempted (保送) due to exceptional or special talent. Why would they exempt special talents from taking the test rather than just giving them special preference beyond the test scores?

    They are pretty transparent about those exemptions. I was trying to get a sample of the students admitted to get the surname distribution before I found out that they were actually lists of those exceptions. There are only two type of exemptions. One is ex-Olympic competitors and they are exclusively sent to Peking and Tsinghua Universities mostly in those elite streams like STEM and Medicine and were given grades of excellent. The others are language specialist trainees for the government or big corporations. Even those sent to the Aeronautical University are language trainees specialized in those areas. They seems to be very serious about specialization. Olympic or any other world competitors will be training full time. Like the World Women Chess champion Hou had said that she was advised not to undertake university study. She ignored that and she admitted that her chess performance was affected.

    Well that sample of language trainees gave me some ideas on the relative importance of some countries relative to China.

    Rank|N|Pct|Language
    1|287|36.5|English
    2|101|12.8|German
    3|91|11.6|Japanese
    4|78|9.9|French
    5|57|7.2|Spanish
    9|19|2.4|Portuguese # Mexico and Caribbean more important than Brazil??
    17|3|0.4|Latin # There are big negotiations with Vatican that upset HK Cardinal Zen.
    24|1|0.1|Tamil # I suppose Indian diplomats speak English
    25|1|0.1|Sanskrit

    The chairman of the Asian Infrastructure Investment Bank AIIB could be formerly one of those trainees. He is still an adjunt professor in those language schools.

    Thank you for all of the additional information! The Monash PDF has an interesting Table 1 showing how much admissions requirements vary by province. For interpreting the following data I think it is worthwhile to know that Monash has significantly lower than average thresholds for admission for Fujian province (489 for full admission). I don’t understand why Shanghai province has the lowest thresholds (except for ethnic Tibetans) with 402 for full admission.

    This paper about gender gaps in the Gaokao has statistical data for 8164 people from Anxi county in Fujian province: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2717464
    Figure 1 has distribution plots and Table 1 has summary statistics. Mean (SD) for the Gaokao in Anxi were 426.34 (80.69) so in theory the ceiling is ~4 SD for that group.

    Read More
    • Replies: @dux.ie
    > I don’t understand why Shanghai province has the lowest thresholds

    Shanghai being a rich region with pop of 23 million has 9 tier 1 and 2 universities, hence the uni entry cut-off is lowered. The neighbouring Zhejiang has 54 million pop and only 2 tier 1 and 2 universities, and thus they are top of the cut-off ratio list. Fujian Hokkien dialect seems to be the main Chinese dialect for Singapore, Taiwan and US EastCoast which might have drained away many of the local students. Similarly for Guangdong with students draining to HK, UK, US WestCoast and Canada. The Monash Gaokao cut-off scores are the exact Chinese provincial Gaokao cut-off scores.

    Incidentally, for those who whinged about the Chinese hukou system it is not as strict as they claimed. Among those top provincial GaoKao students in 2012 a few of them were from boarding schools, i.e. they were most probably from the rural areas. For example the Zhejiang 2012 top GaoKoa Science student, his parents are simple rural farmers from a village in Cixi county with average education and they are not tiger parents. His small town boarding school is in a different Zhenghai county and his junior high results were just slightly above average so he was not those special provincial sponsored students. He has not taken those prep classes. However his results picked up in senior high. In the exam year he moved out of boarding school and his mother joined him to cook and look after him. He scored top in the province with the highest university cut-off score.

    http://translate.google.com/translate?u=http%3A%2F%2Fedu.people.com.cn%2Fn%2F2012%2F0623%2Fc245481-18368523.html&hl=en&langpair=zh-CN%7Cen&prev=%2Flanguage_tools

    > As such, students have an incentive to put in their full effort on the mock exam

    Not so. The authors do not seem to understand competition strategies. In Olympic competitions it is important to time the peak performance window for the actual and final race.

    https://www.tandfonline.com/doi/abs/10.1080/02640410310001655822?src=recsys&journalCode=rjsp20

    "Within competitions, both nations showed similar improvements in mean time from heats through finals (overall 1.2%; 95% confidence limits 1.1 to 1.3%)."

    If you over-stressed yourself in the heats or mock exams, you might not be able to unwind for the final race while still under stressful conditions and handicapped yourself in the real final test. In the Olympic situations you have to exert a bit more to ensure selection for the finals while for CEE that is not important and can be slacker and hence bigger improvements in the finals. There is also the psychological games giving the opponents false sense of competitiveness in the heats or mock exams.

    The sequence of exam subjects also biased against females. In general females are better at languages while men are better at maths and sci. For females the sequence of two consecutive higher stress exams of maths and sci might give them little time to unwind. For the men the higher stress language subjects are separated by two lesser stress subjects might give them more chance to recover.

    http://bjsm.bmj.com/content/bjsports/50/17/1043.full.pdf
    "International Olympic Committee consensus statement on load in sport and risk of illness"

    """It is recommended ... schedule adequate recovery, particularly after competitions, rest, active rest, relaxation strategies"""

    Undertaking lesser stressful exams could be equivalent to "active rest".

    The pop units should be 10,000.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  61. dux.ie says:
    @Jim Bob Lassiter
    Where does Russian place in Chinese foreign language proficiency priorities?

    Rank|N|Pct|Language
    7|48|6.1|Russian
    23|1|0.1|Zulu

    Another interesting data is that Zulu is the only African language prioritized. China gives lots of scholarships to the Africans and I suppose the official communications will be in French, English or Mandarin. I used to know a Senegalese graduate from Peking Uni computer science major and our party trick was to let him ordered Chinese meals for us from the Cantonese speaking waiters who could not understand Mandarin.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  62. dux.ie says:
    @res
    Thank you for all of the additional information! The Monash PDF has an interesting Table 1 showing how much admissions requirements vary by province. For interpreting the following data I think it is worthwhile to know that Monash has significantly lower than average thresholds for admission for Fujian province (489 for full admission). I don't understand why Shanghai province has the lowest thresholds (except for ethnic Tibetans) with 402 for full admission.

    This paper about gender gaps in the Gaokao has statistical data for 8164 people from Anxi county in Fujian province: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2717464
    Figure 1 has distribution plots and Table 1 has summary statistics. Mean (SD) for the Gaokao in Anxi were 426.34 (80.69) so in theory the ceiling is ~4 SD for that group.

    > I don’t understand why Shanghai province has the lowest thresholds

    Shanghai being a rich region with pop of 23 million has 9 tier 1 and 2 universities, hence the uni entry cut-off is lowered. The neighbouring Zhejiang has 54 million pop and only 2 tier 1 and 2 universities, and thus they are top of the cut-off ratio list. Fujian Hokkien dialect seems to be the main Chinese dialect for Singapore, Taiwan and US EastCoast which might have drained away many of the local students. Similarly for Guangdong with students draining to HK, UK, US WestCoast and Canada. The Monash Gaokao cut-off scores are the exact Chinese provincial Gaokao cut-off scores.

    Incidentally, for those who whinged about the Chinese hukou system it is not as strict as they claimed. Among those top provincial GaoKao students in 2012 a few of them were from boarding schools, i.e. they were most probably from the rural areas. For example the Zhejiang 2012 top GaoKoa Science student, his parents are simple rural farmers from a village in Cixi county with average education and they are not tiger parents. His small town boarding school is in a different Zhenghai county and his junior high results were just slightly above average so he was not those special provincial sponsored students. He has not taken those prep classes. However his results picked up in senior high. In the exam year he moved out of boarding school and his mother joined him to cook and look after him. He scored top in the province with the highest university cut-off score.

    http://translate.google.com/translate?u=http%3A%2F%2Fedu.people.com.cn%2Fn%2F2012%2F0623%2Fc245481-18368523.html&hl=en&langpair=zh-CN%7Cen&prev=%2Flanguage_tools

    > As such, students have an incentive to put in their full effort on the mock exam

    Not so. The authors do not seem to understand competition strategies. In Olympic competitions it is important to time the peak performance window for the actual and final race.

    https://www.tandfonline.com/doi/abs/10.1080/02640410310001655822?src=recsys&journalCode=rjsp20

    “Within competitions, both nations showed similar improvements in mean time from heats through finals (overall 1.2%; 95% confidence limits 1.1 to 1.3%).”

    If you over-stressed yourself in the heats or mock exams, you might not be able to unwind for the final race while still under stressful conditions and handicapped yourself in the real final test. In the Olympic situations you have to exert a bit more to ensure selection for the finals while for CEE that is not important and can be slacker and hence bigger improvements in the finals. There is also the psychological games giving the opponents false sense of competitiveness in the heats or mock exams.

    The sequence of exam subjects also biased against females. In general females are better at languages while men are better at maths and sci. For females the sequence of two consecutive higher stress exams of maths and sci might give them little time to unwind. For the men the higher stress language subjects are separated by two lesser stress subjects might give them more chance to recover.

    http://bjsm.bmj.com/content/bjsports/50/17/1043.full.pdf

    “International Olympic Committee consensus statement on load in sport and risk of illness”

    “””It is recommended … schedule adequate recovery, particularly after competitions, rest, active rest, relaxation strategies”””

    Undertaking lesser stressful exams could be equivalent to “active rest”.

    The pop units should be 10,000.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  63. hyperbola says:
    @Factorize
    300k IQ GWAS.

    Is this a new one? Sometimes it is not always clear if it is hot off the press or not.
    https://www.nature.com/articles/s41467-018-04362-x

    Another study that confirms that “cognitive function” is a VERY complex trait at the genetic level – so much so that claims of “prediction” of cognitive function (IQ) by genetic makeup is a fraud.

    Gene-based analyses find 709 genes associated with general cognitive function. Expression levels across the cortex are associated with general cognitive function. Using polygenic scores, up to 4.3% of variance in general cognitive function is predicted in independent samples.

    Given that is so, one wonders when diehards like J. Thompson will finally give up on silly hypotheses about genes/race/IQ correlations that have no validity at the individual level.

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

    hyperbola, my sense is that there is now a large amount of … er … moving of the goal posts as I believe res refers to it. An online article that I read yesterday appeared to hint that a large IQ GWAS is now moving towards publication.

    The article also referred to online services that are already using genotypes as predictors for IQs (though the article notes that such predictions are at least for the time being inaccurate). There has been some critical comments on my fixation on the 1500 IQ people that are on the way. Perhaps I have not focused enough on the much shorter term horizon in which very accurate predictions of children’s IQs will be known.

    How exactly will we cope with a world in which not only could children be quantitatively assigned to genoIQ classes, but such classes would be scientifically provable? Let’s face it, the school-yard taunts are the only insults that actually hurt in life. They arrive at a time in which we do not have the cognitive skills to contextualize or reframe events in their social environment. What happens when the taunts are true?

    With regards to the topic of this thread (for once I’m on topic), why on earth would political activists want children from a particular population to be more fully represented at elite schools, if the level of academic achievement of such children were not competitive? I think it is fair to say that we are now reaching the time in which IQ will become transparent through genotyping. If I were a parent who were thinking of their child’s education as an investment decision, I would certainly not invest hundreds of thousands of pounds (?)(euro, USD) on an elite education if I knew that before graduation a $50 genochip test could accurately predicate their cognitive ability: Cognitive ability is what employers are interested in. If my child did not have a high genoIQ, then investing in an elite education would be a waste of time and money. Yet, so to would be investing in such an education, if my child did have a high genoIQ. It is the prisoner’s dilemma where you can’t win!

    The $50 genetic test could soon be of greater relevance than educational credentials. I am not sure why this idea was not noted in the thread writeup. If it can be proven that these kids at elite schools have these measurable genetic qualities, what would reason would there be for them to be in school?

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter Display All Comments
  65. @James Thompson
    Made an under-estimate of what is covered in "better than AAA". It actually goes to A*A*A*A*A* plus a Subsiduary A. Oxbridge can pick and choose. Not strictly relevant here, but the quality varies by college.

    I simply don’t know enough about the British testing regime. (This CAT test doesn’t even show up in Wikipedia–for CAT i get the common admissions test in India.)

    But obvious question–how comprehensive. The UK has–ballpark–a quarter the population of the US. Our (American) cohorts run 4m per year. That puts Britain at 1m–a quick search suggests that’s pretty spot on. So why is this at a measly 160K? Which 160K is it?

    ~~

    Ok, a bit more poking i found there is something called the Classics Admissions Test, for which scores are required for admission to Oxford. I assume this is it.

    But this will be very self-selected–which explains some of the otherwise incomprehensible numbers (like decent mixed white-black score).

    Tacking on a bit of explanation–what the test is, who takes it–to the post would help.

    Read More
    • Replies: @James Thompson
    Agree that a link would have been helpful.
    https://www.gl-assessment.co.uk/products/cognitive-abilities-test-cat4/
    , @res

    But this will be very self-selected–which explains some of the otherwise incomprehensible numbers (like decent mixed white-black score).
     
    True. That is why we need to compare the test taker numbers with the total population for each group separately.
    , @Edward
    In addition to Dr. Thompson's helpful link, see also my post above. The CAT is taken by around 1 million UK students every year (probably in different year groups) and everyone within a year group takes it. There's no evidence that self-selection is occurring.
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  66. @AnotherDad
    I simply don't know enough about the British testing regime. (This CAT test doesn't even show up in Wikipedia--for CAT i get the common admissions test in India.)

    But obvious question--how comprehensive. The UK has--ballpark--a quarter the population of the US. Our (American) cohorts run 4m per year. That puts Britain at 1m--a quick search suggests that's pretty spot on. So why is this at a measly 160K? Which 160K is it?

    ~~

    Ok, a bit more poking i found there is something called the Classics Admissions Test, for which scores are required for admission to Oxford. I assume this is it.

    But this will be very self-selected--which explains some of the otherwise incomprehensible numbers (like decent mixed white-black score).

    Tacking on a bit of explanation--what the test is, who takes it--to the post would help.
    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  67. res says:
    @AnotherDad
    I simply don't know enough about the British testing regime. (This CAT test doesn't even show up in Wikipedia--for CAT i get the common admissions test in India.)

    But obvious question--how comprehensive. The UK has--ballpark--a quarter the population of the US. Our (American) cohorts run 4m per year. That puts Britain at 1m--a quick search suggests that's pretty spot on. So why is this at a measly 160K? Which 160K is it?

    ~~

    Ok, a bit more poking i found there is something called the Classics Admissions Test, for which scores are required for admission to Oxford. I assume this is it.

    But this will be very self-selected--which explains some of the otherwise incomprehensible numbers (like decent mixed white-black score).

    Tacking on a bit of explanation--what the test is, who takes it--to the post would help.

    But this will be very self-selected–which explains some of the otherwise incomprehensible numbers (like decent mixed white-black score).

    True. That is why we need to compare the test taker numbers with the total population for each group separately.

    Read More
    ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
  68. Edward says:
    @AnotherDad
    I simply don't know enough about the British testing regime. (This CAT test doesn't even show up in Wikipedia--for CAT i get the common admissions test in India.)

    But obvious question--how comprehensive. The UK has--ballpark--a quarter the population of the US. Our (American) cohorts run 4m per year. That puts Britain at 1m--a quick search suggests that's pretty spot on. So why is this at a measly 160K? Which 160K is it?

    ~~

    Ok, a bit more poking i found there is something called the Classics Admissions Test, for which scores are required for admission to Oxford. I assume this is it.

    But this will be very self-selected--which explains some of the otherwise incomprehensible numbers (like decent mixed white-black score).

    Tacking on a bit of explanation--what the test is, who takes it--to the post would help.

    In addition to Dr. Thompson’s helpful link, see also my post above. The CAT is taken by around 1 million UK students every year (probably in different year groups) and everyone within a year group takes it. There’s no evidence that self-selection is occurring.

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

Leave a Reply - Comments on articles more than two weeks old will be judged much more strictly on quality and tone


 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 James Thompson Comments via RSS