I'm reading Dataclysm, a book exploring online data especially related to 23andme.com (huge dating site). There is a chapter on interrace ratings which show the usual patterns, namely that euros(male and female) rate africans lower. They however mention that this isn't the case in the UK. If the theory is that mean g levels cause stereotypes which cause these rating patterns, then perhaps the africans in the UK really are smarter. Unfortunately, they don't share much of the data.

This is unlikely because the fairly representative adult (16 to 65 with mean ages of about 40) samples which I found show a difference of about 0.7 SD. These were from 2000, 2007, and 2012 (see attached, for example), plus two others from 2003 and 2011 which provided numeracy/literacy pass rates. So if there was a substantial narrowing, it was basically with the 2000 birth cohort. And I don't imagine that these kids would have been the referential point in the rating study mentioned.

(There was also an ETS SAT evaluation study in 2008 which showed a 0.7 SD or so difference. In 2012, when the MCS kids were 12 the evaluation study kids would have been in their 20s. So, again, the narrowing would have to have been recent.)

Chuck,

Keep in mind that a growing proportion of "black" people in Britain are actually biracial, i.e., they have one white parent. I've read reports that almost one in two "black" children in London has one white parent. It's difficult to quantify this phenomenon because most of the data is based on self-report. If you consider yourself "black" you're black, even though most of your ancestry is non-African.

Openness to mixed marriage, or simply mixed cohabitation, has greatly increased over time, so the closing of the Black-White IQ gap may simply be an artefact of this demographic change. Another factor is the increase in immigration from the Horn of Africa (Somalia, Eritrea, Ethiopia). These communities are officially categorized as "black" yet they are genetically different from sub-Saharan Africans. I remember reading an article (one of your posts?) which said that the IQ gap between Somalis and British Whites was only 5 points.

Neither of the above factors is trivial.

This is unlikely because the fairly representative adult (16 to 65 with mean ages of about 40) samples which I found show a difference of about 0.7 SD. These were from 2000, 2007, and 2012 (see attached, for example), plus two others from 2003 and 2011 which provided numeracy/literacy pass rates. So if there was a substantial narrowing, it was basically with the 2000 birth cohort. And I don't imagine that these kids would have been the referential point in the rating study mentioned.

It's not implausible.

I've been looking into this topic and here is what I've found.

http://www.independent.co.uk/news/education/education-news/exclusive-must-do-better-black-pupils-did-with-best-improvement-in-exams-9563395.html

Black pupils have achieved the biggest rise in test and exam results of any ethnic group in the country, according to new statistics published tomorrow.

Figures show the gap between their GCSE results and the average for the country has been cut by more than half in just four years – from 5.8 per cent to 2.5 per cent last year.

Nationally, 58.1 per cent of black pupils achieved the benchmark of five A* to C grade passes at GCSE including maths and English – up 8.8 per cent since 2010.

The improvement is most marked amongst the poorest black boys who have reduced the gap in the percentage reaching that benchmark by 4.4 percentage points since 2009 – bringing the total achieving the benchmark up to 43.1 per cent.

The figures mark the end of a long period during which black pupils have languished at the bottom of the league table detailing the performance record of different ethnic groups in exams.

In both English and maths, black pupils outscore the rest of the country at secondary school in terms of the progress made – with 76.2 per cent and 74.2 per cent of pupils making the expected level of progress or better – compared with 70.4 per cent and 70.7 per cent for all pupils.

Key reasons for the improvement include the introduction of the English Baccalaureate – which has persuaded schools to enter thousands more pupils for core academic subjects since its introduction in 2011. Under it, schools are ranked in league tables on the percentage of pupils getting top grade passes in maths, English, a language, a humanities subject and the sciences.

In addition, schools have been able to earmark “pupil premium” funding, given to them for every disadvantaged pupil they take, to ensure poorer pupils catch up – and are then able to access the more academic subjects.

The progress is not limited to the GCSE. From the same article:

The success story also holds for the results of national curriculum tests for 11-year-olds in maths, reading and writing – where 73 per cent of black pupils reached the expected benchmark last year – only two percentage points behind the national average compared with a five percentage point difference in 2010.

Schools minister Lord Nash said: “For years black pupils’ results have lagged behind their peers but that gap is being eroded at all levels – The Government’s school reforms are helping thousands more black pupils, including the poorest, to do well at primary school, thrive in their GCSEs and succeed in life.”

Those would be the same age as the kids in the Millennium Cohort Study...

The increase in achievement has prompted major changes in the way some charities approach helping disadvantaged young people. For instance, Generating Genius, launched by black academic Dr Tony Sewell several years ago to improve the performance of Afro-Caribbean young people in the sciences, has now widened its target audience to take in disadvantaged young people of all backgrounds and is starting a new project in the Medway Towns in Kent.

More info about this program:

http://www.independent.co.uk/news/education/schools/generating-genius-began-tutoring-gifted-afrocaribbean-boys-in-science--now-the-charity-is-mentoring-boys-and-girls-of-all-races-9413540.html

In 2005, 10 gifted Afro-Caribbean boys were tutored in science to try to reverse the trend for under-achievement in their ethnic group. Now, Generating Genius mentors 900 boys and girls

It started off as a project to help what was then the UK's most under-performing group of pupils – Afro-Caribbean boys – to succeed in science, an area that they might never have thought of studying. Generating Genius, a charity founded by respected academic Dr Tony Sewell, selected 10 boys and set about teaching them, with the aim of helping them to win places at top Russell Group universities.

There was great competition for the places on the scheme, which was advertised in the black community newspaper, The Voice. In all, 200 young people applied for the 10 places. The number on the project soon expanded to 40.

"We looked at the figures for Afro-Caribbean boys at that time [2005]," Dr Sewell says. "Under every indices in terms of exclusion, they were the under-performers. I wanted to show to them, their parents and the world that under-achievement was something that we could easily do something about."

The 12- to 13-year-olds on the programme were selected because they were top performers in their national curriculum tests at 11 – the kind of pupils who should have gone on to get A* or A grades at GCSE, but frequently did not.

For the first 10 pupils, the course began with a four-week trip to Jamaica. "Jamaica is not associated with science, but it has a good university in the University of the West Indies," Dr Sewell says. "We wanted them to be in an environment where they could see scientists, policemen, prime ministers and university vice-chancellors who looked like them and had obviously succeeded."

The project caught the attention of the media both in the UK and in Jamaica – as a result of which the boys were interviewed as they touched down to start their four-week stay.

"They became celebrities," says Dr Sewell. "They had people pointing at them in the street and saying: 'They're the geniuses!' We were trying to create a cohort of bright kids for whom it felt safe to actually be bright. On their estates, it was not actually cool to be bright. We have these terms, 'geek' and so on, to describe people who do well in science."

The project did not stop when they returned to the UK. The boys were assigned mentors to help them with their studies. This is where Dr Sewell believes his project differs from others, which may offer pupils a glimpse of what it might be like to succeed, but then do not stay with them during the five years of compulsory schooling to help them through to GCSEs.

It certainly seems to have paid off. Of those first 40 boys, 38 succeeded in getting a place at a Russell Group university (which represents 24 of the country's most competitive universities, including Oxford and Cambridge) to study a science.

"In addition, we had two who went to the London School of Economics to study economics," Dr Sewell says. It was a 100 per cent success rate.

Not surprisingly, demand for the project increased and Generating Genius expanded its horizons to take in girls as well. The president of the National Union of Teachers, Max Hyde, herself a science teacher, indicated in an interview in The Independent last month that she intended to spend part of her presidential year campaigning against gender stereotyping and for more girls to pursue careers in the sciences.

In addition, research began to show that it was white working-class boys who were now the most poorly performing ethnic group in the UK, rather than the Afro-Caribbean boys.

If I recall correctly, Jim Flynn et al. claimed that the failure of Head Start to create lasting IQ gains after the program ended was due to the fact that the program ended without any follow-up. The program above takes a different approach by assigning them mentors who remain with the students throughout their secondary education. Maybe that's what matters?

Somalis seem to have hired private tutors as well:

http://www.economist.com/news/britain/21583710-somalis-fare-much-worse-other-immigrants-what-holds-them-back-road-long

Many are hopeful. Somalis want their children to succeed, so growing numbers are hiring private tutors (see article). In 2000 just one Somali teenager in the London borough of Camden passed five GCSEs with good grades. To improve matters, the council and others set up the Somali Youth Development Resource Centre, which mentors students and lends them books. Last year the figure rose to 59%.

Abdikadir Ahmed, who works there, says his organisation encourages people to put the entrepreneurial skills they learn in gangs to better use. He works with Somalis locked up in Feltham prison, a young-offenders jail. Their numbers are dropping, he reckons. Somalis played little part in the summer riots of 2011.

Sam Freedman also noticed the improvement in Black GCSE scores earlier this year on his blog:

http://samfreedman1.blogspot.co.uk/2014/02/why-are-black-pupils-doing-so-well.html

Yesterday I was playing around with GCSE data and found an interesting trend. Over the past five years black pupils have almost caught up with white pupils in the main 5 A*-C measure including English and Maths.



My first thought was that this must be a function of the "London effect" as over 60% of black pupils live in the capital. But actually black improvement has been faster in other regions. (Health warning on this chart - some regions have very low numbers of black pupils, especially the North East where just 139 took GCSEs in 2013).



My next thought was that it might be a result of immigration. Unfortunately we don't know how many recent immigrants are included in the figures but we can look at differences between black Caribbean and black African as a proxy, as the latter are more likely to be recent immigrants.

It turns out that the percentage of black African pupils as a proportion of all black pupils has increased over the past 5 years from 54% to 59% and black Caribbean has decreased from 33% to 29%. This explains some of the faster rate of improvement for black pupils overall as black Africans do about 8 percentage points better than black Caribbeans.

But the rate of improvement has actually been faster for black Caribbeans (13.9 percentage points over the past five years compared to 12.8 for black Africans). So there's something else going on as well...

I then noticed that black boys have improved faster than black girls over the past five years (14.4 percentage points to 12.8). This is the opposite of what's happening with white British boys and girls (7.8 to 11.2). It seems like this gender difference is another part of the jigsaw.

Then I dug into the gender/ethnicity data a bit more and found that white British boys not black boys were the outliers.





Despite all the attention on the under-performance of white British boys this surprised me. I had thought the issue was with poor white British boys. But white British boys on Free School Meals make up a small proportion of these figures (and are improving at the same rate as their non-FSM peers - albeit from a much lower base).

So to answer my initial question I've identified two reasons why black pupils are improving faster than white ones. First there are a larger proportion of higher performing black Africans. And secondly black boys are improving faster than white British boys (as are Pakistani/Bangladeshi boys).

I don't think grade inflation explains it - white boys should benefit from this as much as any other group. Also the FSM gap has been pretty static for the past few years.

The figures mark the end of a long period during which black pupils have languished at the bottom of the league table detailing the performance record of different ethnic groups in exams. In both English and maths, black pupils outscore the rest of the country at secondary school in terms of the progress made – with 76.2 per cent and 74.2 per cent of pupils making the expected level of progress or better – compared with 70.4 per cent and 70.7 per cent for all pupils.

It's hard for me to say since I don't have a good grasp of the psychometric characteristics of the exams. The IQ test data itself is not unequivocal. The large CAT normative study which the folks at GL assessment confirmed for me showed an average ability difference of about 0.5 SD. The composite difference would have been a little higher at e.g ~ 0.55 SD. That was for age 11/12 in year 2009-2010. The total Black sample size was about 4500 or about 10x the size of the MCS one. I was informed that it was a representative sample. So, where one to do a meta-analysis using just these two points one would get an age 11-12 year 2009 to 2012 B/W difference of 0.5 SD. Generally, the CAT sample should be taken as a better index since it involved 9 subtests, not just one.

My first thought was that this must be a function of the "London effect" as over 60% of black pupils live in the capital. [b]But actually black improvement has been faster in other regions.

According to Steven Strand, U.K. Black kids are concentrated in a few dozen schools. This would allow schools to focus resources to boost scores. I suspect that a large percent of the variance in the GCSE exams is non-g related; I can't be sure though.

My next thought was that it might be a result of immigration. Unfortunately we don't know how many recent immigrants are included in the figures but we can look at differences between black Caribbean and black African as a proxy, as the latter are more likely to be recent immigrants.

What you would want to do is determine with a high degree of accuracy what the ~ age 12 second+ generation West Indian B/U.k. W gap was from the 1970s to 1990s. Lynn's estimates can't be taken at face value. I have most of the studies if you want to give it a try. Anyways, this would give you a baseline. I looked through the data and I would tentatively put it at 0.8 SD, but it really should be thoroughly reviewed.

I don't think grade inflation explains it - white boys should benefit from this as much as any other group.

It would depend on the amount of inflation. So, here's an example. d is the standardized difference, G1 and G2 are equivalent percentiles. Percentile difference is the percentile difference. As grades inflate, the percentile differences become smaller; when they are smaller they are over sensitive to other influences in the sense that you don't have to shift the percentile distributions as much to "close gaps".

SD G1 G2 Percentile Difference
0.50 0.50 0.31 0.19
0.50 0.75 0.57 0.18
0.48 0.85 0.71 0.14
0.51 0.90 0.78 0.12
0.52 0.95 0.87 0.08
0.51 0.99 0.97 0.03

Because of this, GCSE 5 A-G pass rate differences, with the 95% White pass rates, strike me as being pretty unreliable as indexes of mean differences. But the 5 A-C ones should have discriminatory power. So one would need another explanation.

I don't think -- and I haven't argued here -- that the UK educational establishment closed the UK B/W IQ gap. It does seem that it has been substantially narrowed, though, such to allow non-g factors to make up for cognitive differences when it comes to achievement tests.

Sam Freedman also noticed the improvement in Black GCSE scores earlier this year on his blog:

What's interesting about the Sam Freedman graph is how small the differences were from the start.

Year /White % rate /Black % rate /B/W d
2008 / 0.51 / 0.45 / 0.15
2009 / 0.55 / 0.5 /0.13
2010 /0.58 / 0.55 /0.08
2011 /0.585/ 0.55 /0.09
2012 /0.6/ 0.575 /0.06

Year d/ 0.23/ 0.31

Can you find scores for 2000 to 2008? Or 1990 to 2000?

I would like to see the decade trend.

[quote][/quote]

You wouldn't be interested in writing a paper on this topic, would you? We probably could publish it in Intelligence or if not then here. Something interesting seems to be going on in the U.K. It definitely deserves more investigation.

No one has addressed either of my criticisms:

1. A growing proportion of "Black Caribbean" students are actually biracial. This factor in itself would lead to a narrowing of the IQ gap over time.

2. The category "Black African" includes a growing proportion of immigrants from the Horn of Africa: Ethiopia, Somalia, and Eritrea. Their history has been very different from that of sub-Saharan Africans, and any process of gene-culture co-evolution would have likewise been different.

Answers, please.

No one has addressed either of my criticisms:

1. A growing proportion of "Black Caribbean" students are actually biracial. This factor in itself would lead to a narrowing of the IQ gap over time.

2. The category "Black African" includes a growing proportion of immigrants from the Horn of Africa: Ethiopia, Somalia, and Eritrea. Their history has been very different from that of sub-Saharan Africans, and any process of gene-culture co-evolution would have likewise been different.

Answers, please.

I actually looked into both of these issues. As for (1), the MCS has variables for parents' race. So you can just do a cross tabs analysis. Generally, most kids (~90%) that were called Black had two self identifying Black parents. And most kids (~90%) that were called mixed race (Black-White) had one self identifying Black (or mulatto) and one self identifying White (or mulatto) parent. As for (2), you can just restrict consideration to West Indians, who are 2nd to 3rd generation -- and for whom we have an idea about selection and about score differences. Lynn did a poor job on his review of race differences in IQ in the UK. But he wasn't too far off. You can see a fair review of early West Indians/White British studies here:

Pumfrey, P. D., & Verma, G. K. (Eds.). (1990). Race relations and urban education: contexts and promising practices. Psychology Press.
Chapter 15: reading attainment and examination results of British west indian pupils: challenges and responses to underachievement

I don't think that the 70s to 90s (second+ generation) gap was more that 0.8 SD -- because the three nationally representative samples -- two discussed in the Swann report -- and one discussed elsewhere don't, on average, put it above 0.7 SD. And 0.7 SD is what I get for adults, who would have been these kids 40 years ago. For comparison, Lynn (2008) put it at .93 SD.

But this really should be reviewed. Along with differences in GCSE say in the 80s and 90s.

It's hard for me to say since I don't have a good grasp of the psychometric characteristics of the exams. The IQ test data itself is not unequivocal. The large CAT normative study which the folks at GL assessment confirmed for me showed an average ability difference of about 0.5 SD. The composite difference would have been a little higher at e.g ~ 0.55 SD. That was for age 11/12 in year 2009-2010. The total Black sample size was about 4500 or about 10x the size of the MCS one. I was informed that it was a representative sample. So, where one to do a meta-analysis using just these two points one would get an age 11-12 year 2009 to 2012 B/W difference of 0.5 SD. Generally, the CAT sample should be taken as a better index since it involved 9 subtests, not just one.

I presume you're referring to this?



What struck me the most about those results was the small gap between Whites and Mixed Whites/Black Caribbeans (this should also alleviate some of Peter Frost's concerns about biracials).

This old GNXP post discusses the studies of Muttarak on intermarriage in the UK and here is what it notes:

http://www.gnxp.com/MT2/archives/003835.html

It will be seen that for Black Caribbeans and Chinese the proportion intermarrying has increased since 1981. For Indians it has increased very slightly, with a fall in the rate for men offset by a larger rise in the rate for women. Among Pakistanis and Bangladeshis there has also been a fall among men and a rise among women, with the net effect a fall. Among Black Africans the rate among both men and women has fallen.

...

Muttarak (2) notes that for Black Caribbeans having a higher education qualification makes little or no difference to the propensity to intermarry, or is even a negative factor. She suggests (p.26) that this is because Black Caribbeans are integrated into the white working-class community. The same appears to be true for second-generation Black Africans.

Since a very large percentage of the Black Caribbean/White marriages involve working class folks with limited education, it's hard to claim that the small gap on the CAT is the result of assortative mating for 'g'. And the N value for Mixed Whites/Black Caribbeans is so robust (1589!) that it can't be shrugged off as an anomaly or a sampling error. And, as you say, the CAT was comprised of numerous subtests.

I think that's very hard to reconcile with a large genotypic gap, especially in light of other trends in the UK. Perhaps a modest one (.5 SD?) at most.

Let me get back to you on the rest of your queries.

I think that's very hard to reconcile with a large genotypic gap, especially in light of other trends in the UK. Perhaps a modest one (.5 SD?) at most.

I reviewed a bunch of studies earlier. Regarding the CAT study, I noted: "You will notice there that the UK Black-White gap is only about 0.5 SD. (Using the first regression formula presented in Wright and Strand (2005) –(.66V + .21Q + .18NV) — table 4, the FSIQ gap is 8.2 points.) And the UK Mixed-White FSIQ Gap is only 30% of that." The 30% (of the overall d value) is actually consistent with mixed race scores in the U.S. in that these scores also tend to be closer to the White mean.

Regarding the statement, "I think that's very hard to reconcile with a large genotypic gap", you really must start by considering what a large genotypic gap would minimally predict, holding environmental factors constant. I think a large additive genetic risk score, e.g., d = 1, between Europeans and West Africans is tenable. The Black West Indian population in the UK would be about 85% African. Based on the MCS genealogical data, mixed race B (Caribbean) -W would be about 40% African. Now, from what I have read, West Indian Black UK immigrants were somewhat selected (relative to stayers), though nothing compared to African ones. But let's assume no selectivity. We would have UK BWI/W genotypic d 0.85 and UK White-BWI/W genotypic d 0.40. Now let's assume a heritability for the cognitive tests mentioned of 0.75 for adults and 0.5 for adolescents. We would predict mean differences of, respectively, Adult B/W SQRT = (0.75)*0.85 = 0.74, Adolescent B/W = SQRT(0.50)*0.85 = 0.60, Adolescent mixed W-B/W = SQRT(0.50)*0.40 = 0.28. The adult gaps are consistent; the adolescent gaps are lower, but not horribly so.

Anyways, what we really need is more data and the best way to get that would be to write up a paper on this and see if we can elicit responses -- if you want we could write it critical of a racial hereditarian hypothesis. If you're uninterested in working on a paper, I will be less interested in continuing this discussion with you.

Chuck,

All of these race designations are based on self-report, and when you use self-report there is a strong tendency for biracials to identify as "black." In the UK, "blacks" are increasingly biracial people who are more black than white, and "biracials" are biracial people who are much more white than black, i.e., they have visible African ancestry, but not enough to be considered "black."

This phenomenon was discussed in a recent BBC news article:

"As with the census and other surveys, ethnicity is defined in the UKHLS by the individual: if you regard yourself as black Caribbean or white British that is how you are counted.

Using this self-reported approach, the figures suggest that 0.88% of adults define themselves as "mixed".

But the survey - following 100,000 people in 40,000 households - asks another question: what is the ethnicity of your parents?

The footnote puts it: "If we use this alternative definition of mixed then 1.99% of adults are of mixed parentage."

More than twice as many over-16-year-olds are technically mixed race than describe themselves that way.

Underestimated

Self-definition, of course, also applies to under-16s (parents will normally described the ethnicity of their children) and this group accounts for half of the mixed race population.

There is research evidence which suggests the number of mixed-ethnicity children is also significantly larger than the official figures show.

Self-reported data show 2.9% of children described as mixed race. But the proportion of children living with parents from different ethnic groups or in a mixed-race household is shown to be 8.9%.

Further support for the contention that the number of mixed-race children is under-counted emerges from work on single parents.

The proportion of children in lone-parent households who are of a different ethnicity to the single mum or dad is 8%."

http://www.bbc.com/news/uk-15164970

"...and when you use self-report there is a strong tendency for biracials to identify as "black." In the UK, "blacks" are increasingly biracial people who are more black than white, and "biracials" are biracial people who are much more white than black, i.e., they have visible African ancestry, but not enough to be considered "black.""

This is more true in the US than elsewhere and should not be assumed of the Uk where the one drop rule has never existed. It may in fact be more likely(or as likely) that biracials "much more white than black" identify as White or "English" as the article below describes.

http://www.economist.com/news/britain/21595908-rapid-rise-mixed-race-britain-changing-neighbourhoodsand-perplexing

"Rob Ford of Manchester University points out that Caribbean folk are following an Irish pattern of integration, in that their partners are often working-class. The Irish parallel also suggests they will eventually be fully absorbed into the British population. Polls show that adults who are a mixture of white and black Caribbean tend to see themselves not so much as black, Caribbean or even as British, but rather as English—the identity of the comfortably assimilated."

Emil,

About half of Black Caribbean men in the UK are partnered with a woman from another ethnic background. It's very likely that half of "black" children in the UK are at least half-white by ancestry:

"Across the other ethnic groups, inter-ethnic partnerships were relatively rare, with the exception of the Black Caribbean group where nearly half of men in a partnership were partnered with a woman of a different ethnic group."
http://www.equalityhumanrights.com/sites/default/files/documents/raceinbritain/ethnicity_and_family_report.pdf

Jm8,

Please, I wasn't "assuming" anything. I provided a reference with supporting statistics. There is a strong tendency in the UK to classify biracial individuals as "black." This isn't just a strange American tendency. Keep in mind that the "mixed" census category is recent and goes back only to 2001. Previously, UK biracial individuals were counted as "black."

Emil,

About half of Black Caribbean men in the UK are partnered with a woman from another ethnic background. It's very likely that half of "black" children in the UK are at least half-white by ancestry:

"Across the other ethnic groups, inter-ethnic partnerships were relatively rare, with the exception of the Black Caribbean group where nearly half of men in a partnership were partnered with a woman of a different ethnic group."
http://www.equalityhumanrights.com/sites/default/files/documents/raceinbritain/ethnicity_and_family_report.pdf

Peter,

I don't see why your point is germane. Take the GL assessment report. The 1,978 West Indian "Black" kids perform 0.5 SD below the UK White kids, while the 1,589 West Indian+White kids perform about 0.20 SD below. Here, about half of the individuals with some Black ancestry are identified as biracial. Now, perhaps some of the mixed race kids are being identified as Blacks; if so, the Black scores would be inflated. But this leaves the overly small mixed race versus White difference. The point is that the (Black + 2 x Mixed race) gap is overly small. Do the math.

Chuck,

My point applies just as much to the "mixed" group as to the "black" group. A "mixed" person is someone who is predominantly white but has some visible African ancestry. Mariah Carey is often cited as a typical "biracial" person, yet she is only 1/4 black. If a person is 1/2 black, there is a high probability that he or she will simply be considered "black."