[Update: Kiwiguy dug up 2006-2010 UKCAT annual report scores. I added them to the discussion below. I also added 2008-2009 UK laws school test (LNAT) results]
As usual, JL managed to unearth some obscure document as counter-weight to my otherwise solid case against “race realism.” Below, I summarized the data. Pertinent are the last couple of rows, where White, Mixed, and Black IQs are presented, normed on a White mean of 100 with Standard deviations of 15. 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.
Wright and Strand (2005) give some background on CAT3.
The CAT-3 is now the most widely used test of reasoning abilities in the UK, used routinely in educational settings, contributing to over 850,000 pupil assessments in 2002/2003. The test battery comprises 3 scales measuring Verbal Reasoning (VR), Non Verbal Reasoning (NVR), and Quantitative Reasoning (QR) and is traditionally administered in a group classroom setting. Children are provided with answer booklets and are given extensive introductory instructions and practice examples as to how to approach each subtest in the battery. A large scale study carried out by Strand (in press) using CAT-2 (a precursor to the CAT-3), reported very high test-retest correlations over a three-year interval in children aged 10 to 13 years. Correlations ranged from 0.76 for NVR, 0.79 for QR, 0.87 for VR, and 0.86 for mean CAT-2 score. CAT scores are strongly correlated with pupils’ subsequent attainment in national tests at age 11 and 14 and public examinations at age 16-18 (Smith, Fernandes, & Strand, 2001). According to data provided by nferNelson, most children sit the CAT-3 test in year 7 (age 11-12 years) and around 20% of all children will have CAT-3 tests in both year 7, and year 9 (age 13-14 years) (S. Strand, personal communication, September 2004). Many children in primary schools will take elements from CAT-3 rather than the full battery. The widespread use of the CAT-3 as a formal test of children’s intellectual abilities therefore provides a potential method of estimating children’s premorbid IQ. (Estimation of Premorbid Intellectual Abilities in Children)
I find the data ambiguous. The kids are only ages 11-12 here. As made clear in a figure which I pointed to previously, the White-Black African GCSE point gap decreases with age, from which we might infer that the IQ gap decreases with age.
I was also able to find a tiny sample (n=11) coming from a longitudinal study conducted in 2000. Age 11 verbal reasoning test scores were reported. These kids would have been born around 1983. As can be seen, the difference is .78 SD.
Frederickson and Petrides, 2005. Ethnic, gender, and socio-economic group differences in academic performance and secondary school selection: A longitudinal analysis
More data comes from wave 4 of the the nationally representative Millennium Cohort Study. The kids were tested in 2008 with the British Ability Scale. The children in this wave of the study are 7. The first column shows the gQ score, the second column shows the standard error, and the last shows the sample sizes. The Black-White difference, here, is a mere .33 SD. For comparison, Flynn and Dickens (2005) give an age 7 US difference of .5 SD.
I was able to find some UK SAT data based on year 13 (age 17) students. The kids took the tests in 2006:
What’s noticeable (Table 5.5 page 30) is that the SAT and GSCE score gaps match; we might infer then that since as of 2010 there is little GSCE gap, there would be little SAT gap. But the sample sizes are small here and probably unrepresentative, so it’s difficult to make much of this.
What’s needed then are CAT scores for adults. The only ones which I found, dating from 2000, show such outlandishly high gaps, that they are hardly credible indexes of 2nd generation adult Black performance. My interpretation would be that these represent the scores of mostly first generation immigrant Blacks.
The magnitude of these gaps do more or less match with those found based on Situational Judgment tests (STTs). With SJTs, test-takers are presented with realistic scenarios and asked to identify the appropriate response. These tests are found to show less mean ethnic/racial differences than traditional cognitive ability tests. (See: Whetzel, et al., 2008. Differences in Situational Judgment Test Performance: A Meta-Analysis.)
Source: Dewburry et al. Situational judgment tests and adverse impact in the UK: A large sample study
Other data comes from the 2012 Programme for the International Assessment of Adult Competencies survey in which the Literacy and Numeracy of age 16 to 65 adults was measured. On page 90 of tthe report for England, The international survey of adult skills 2012: adult literacy, numeracy and problem solving skills in England, the following Black/White score differences were provided:
On page 15 of the same paper, the following sample sizes are reported:
On pages 259 thru 266 of the OECD Skills Outlook 2013 report, the following unadjusted English SDs are reported:
Literacy 49.1 (pg 261)
Numeracy 55.0 (pg 266)
The correlation between Literacy and Numeracy in England was given on page 266:
r (L x N) = .87
Based on the above, the composite score would be:
0.737 / ((SQRT (2+2*0.87))/2) = 0.76.
Similar results were found based on the reported 2003 and 2011 English Skills for Life surveys when restricting consideration to UK Whites and UK Caribbean Blacks.
Some other data comes from a decade ago:
You will notice here that the gaps are only 0.5 SD; see also table 6.2. These samples are likely unreliable as the sizes are so minute.
A commenter pointed me to UKCAT annual report scores; the UKCAT is a dental and medical school entrance exam. The test is said to be a measure of aptitude and not knowledge (UKCAT Annual Report, 2007. pg. 2). It measures Verbal Reasoning (VR), Quantitative Reasoning (QR), Abstract Reasoning (AR), and Decision Analysis (DA). I transformed the total scores into UKCATIQ scores, normed on a White mean of 100 with standard deviations of 15. The year, sample size, and scores are listed:
The results from the most recent report — the 2010 to 2011 annual report — can be seen below. Based on the information presented in the annual reports the average age of the test takers was 21. So these individuals would have been born in 1990.
Ethnicity………….N…….% of test takers……..CATIQ
It can be seen that there are consistent score differences. On these exams, Blacks score approximately 1 SD below Whites. Medical school applicants are not representative of the population at large, though, so it’s not clear how much analytic leverage this data gives us.
Other medical tests, though, do show similar differences. Here are the results from a recent pilot study of the Clinical Problem Solving Test — with higher rank scores representing worse results:
Dacre, et al., 2008. Clinical Problem Solving Test Pilot Project Report.
I was able to find LNAT test results from 2005 to 2009. The LNAT is the UK law school test. I transformed the total scores into LNATIQ scores, normed on a White mean of 100 with standard deviations of 15. Standard deviations were not provided, but the data allowed for their calculation. I calculated them for the total sample. The White SDs are lower, so the actual standardized difference between Blacks and Whites, based on the pooled White-Black SD, is higher than shown below. The first table below lists the results in detail for the 2008-2009 cycle; the second summarizes the results from 2005-2009.
LNAT: National Admissions test for Law. 2010. Analysis of LNAT results 2008-9 by gender, ethnicity, school and parental occupation. Accessed February 2012.
This data agrees with the Bar rankings. The qualification rank differences comes out to between 0.5 and 1 SD.
Sauboorah, 2011. An analysis of full-time students enrolled on the 2009/10 BVC See also: Dewberry, 2001. Performance disparities between whites and ethnic minorities: Real diVerences or assessment bias?
Also, adverse impact scores were provided for the UK Fire Fighter tests, the NFA. For the subsections of this test that were cognitively loaded the White-non-White gap was 0.9 SD. The sample sizes were small and the “non-White” category was not broken down by ethnicity. But we can reasonably conclude that the scores of Black and South Asian test takers were well below those of the White ones, given the composition of the UK non-White population at large. This study was conducted in 2009 and the median age of the participants, deduced from the second source cited, was about 25. So these individuals would have been born in 1985.
I did find some data on university attainment. Averaging across degree classes, Whites obtain approximately 0.5 standard deviations higher degree classes than Blacks. But it will be noticed, based on the Chinese sample, that there doesn’t seem to be a horribly good association between IQ differences and attainment differences:
Source: Richardson (2008). The attainment of ethnic minority students in UK higher education. See also: Richardson, 2010. Widening participation without widening attainment: The case of ethnic minority students
Overal, to my mind, the data is ambiguous. What is noticeable is that to the extent the CAT data captures the UK Black-White difference, the gap has greatly narrowing in the last couple of decades. Compare here. Based on the data Lynn provided I calculated a 0.9 SD difference. So the gap possibly narrowed 0.4 SD. This is rather amazing since, undoubtably Black immigrants to the UK have become much less selected, as the government now virtually sweeps African Blacks out of refugee camps into the UK.
When I get a chance, I will look into UK immigrant selection.