If you missed it, I referenced Jensen’s latest paper below. I guess he hasn’t got word of my Hereditarian debunking.
(Hat tip Human Stupidity)
Perplexed, Human stupidity states: “Amazing how white and black Americans have totally different perceptions. The truth can only be one, and the same evidence is known to all.”
Amazing, indeed. And yet as Poincare once noted, it is the job of scientists (and, we might add, internet bloggers) to identify the regularities which tie the many amazing curiosities of our world together.
The standardized difference in belief in guilt is 1.1 (0.72 to 0.32), the standardized difference in belief in innocence is .85 (0.01 to 0.07), and the standardized difference in agnosticism is .89 (0.27 to 0.61). Where have we seen those magnitudes before? To alleviate your curiosity, refer to page 254-269 in Robert Gordon’s magnum opus, Everyday Life as an Intelligence Test: Effects of Intelligence and Intelligence Context.
72% of Blacks, 31% Non-Blacks Believe George Zimmerman is Guilty: Why the Racial Differences?
Blacks are much more likely than nonblacks to have an opinion about Zimmerman’s guilt. Overall, 72% of blacks say Zimmerman is definitely or probably guilty of a crime; 1% say he is not. Nonblacks also say Zimmerman is guilty, by 32% to 7%, but well over half of nonblacks say Zimmerman’s guilt is unclear from the available information.
Blacks are more certain about their opinions than are nonblacks. Blacks who say Zimmerman is guilty of a crime are significantly more likely to say he is definitely guilty than probably guilty, while nonblacks tilt more toward the “probably guilty” choice.
Additionally, 72% of blacks say racial bias was a major factor in the events that led up to the shooting death of Martin, with another 13% saying it was a minor factor. Nonblacks, on the other hand, are significantly less certain, with 31% saying racial bias was a major factor, 26% saying it was a minor factor, and 25% saying it was not a factor at all Gallup
To better understand the UK data, I plotted the standardized difference by birth year, which I simply estimated as study publication date minus median subject ages. The studies presented by Lynn (2006) are in red; the studies that I have found are in blue. As for the latter, I had the guesstimate subject ages for a few samples. For the clinical and law aptitude tests, I assigned an age of 30, since in one of the data sets this was the median age of the applicants. For the industrial samples, I assigned an age of 40, since this is the median age of workers. I also correct the Situation Judgement Test difference up to 1, as in the US a difference of 0.5 on these tests corresponds to a cognitive differences of 1 (refer to Roth, et al. for the intercorrelations). As For the military samples, I assigned an age of 20, on the assumption that this was the median age of armed force applicants. I also had to estimate the standardized difference based on SAT scores, which I did by assuming a White standard deviation of 100. It should be noted that all of the studies were problematic in some way. They were either based on convenience samples (e.g., most of Lynn’s samples), based on unrepresentative samples (e.g., the clinical and law samples), unpublished (e.g., the gl-assessment sample), based on subtests scores and not full scale IQs (e.g., the MCS sample), and so on. Generally, the gaps shows a secular narrowing, but are not particularly consistent with the hypothesis that the cognitive ability gap has closed in the last generation.
I also plotted the standardized difference by test year. This, it can be seen, paints a somewhat different picture. This is due to my data points having a number of adult samples. Contemporaneous British Blacks age 30 and up, it seems, score around 0.8 standard deviations below Contemporaneous British Whites. An important unresolved question is whether or not the gap increases with age longitudinally as it does in the US. The narrowing shown when plotting data by birth date could represent a narrowing merely at young ages, for a given birth cohort, when the heritability of IQ is low, or it could represent a genuine narrowing of the population difference across age cohorts. The former would be consistent with a genetic hypothesis, while not the latter.
This then leaves a puzzle. The median birth year of the various achievement samples, which show trivial to small gaps, is 1992. (The kids were 14-16, median 15, and the tests were taken between 2004 and 2010, median 2007.) Based on the trend line from the cognitive data below, the IQ gap should be 0.7 SD for this cohort. Given the correlation found between achievement scores and IQ (0.7), the minimum that the achievement gap should be, given this IQ difference, is .5 SD (0.7 x 0.7). Or conversely, given the achievement gap of maybe 0.2 SD — averaging the Black Caribbean and Black African scores — the most the cognitive gap should be is .3 SD (0.2/0.7). Why the disagreement in data? This, I suggest, is a question which warrants further investigation. Something is amiss in the UK.
There is a second puzzle here, too. The data shows that contemporaneous British Blacks age 30 and up are deficit in cognitive ability around 0.8 standardized units relative to Whites. The data implies that this magnitude of differences has been there for 40 years since testing began. If, for this last generation, the gap is greatly reduced, when did the narrowing begin? What is the magnitude of the gap for contemporaneous British Blacks age 20 and up?
I just can across this discussion on Bruce Charlton’s Miscellany. Charlton cites a paper by Irwin Silverman which shows a secular increase in simple reaction time. I plotted the data points below. Different methods were used across studies, so there’s a lack of method invariance and the results are not compelling. But they agree with other findings in showing no ECT Flynn effect. (Increased intelligence should correspond with decreased reaction time.) The results are interesting for three reasons: (1) ECTs are based on a ratio scale, which more readily allows for cross temporal and spatial comparisons (see: Jensen (2011)’s discussion of the Flynn effect in this respect); (2) the correlation between ETC and IQ is largely genetic;（3）and relatively consistent racial differences in ECTs have been found.
If you are not familiar with the subject, you can read Jensen’s 2006 book on the topic. PDF, for your reading pleasure, here: Jensen, A. R. (2006). Clocking the mind: Mental chronometry and individual differences. Elsevier Science.
If any readers are interested, I would be willing to pay for an analysis of the UK 2006/2009 PISA data by ethnicity by nativity. (Were you to do this as a charitable service, I would be much obliged — but I appreciate how adverse HBDers are to these types of activities.) You will need access to a statistical software package such as SPSS and some familiarity with it. (If you need help, I can walk you through the process.) I am only looking for very simple trivariate analyses, such as the one I did here. You can download the data online. I would do it myself but I’m currently short a few moles of ATP (and, due to my computer’s untimely death — I use my wife’s now — the requisite software). If you are interested let me know. Thanks.
[Update: Random Human left a comment which sums up my feelings on this matter:
Yeah, the tests seem to discriminate just fine. This is really troubling data for the racial-hereditarian position and has caused me to update away from it to a degree. It’s obviously not a slam dunk that completely does away with the debate but it’s really troubling data. Even more troubling is the “meh, whatever” reaction from the top hereditarians. Reasonable objections have been raised by the commentators here but no-one has really succeeded in explaining this away.
In response to my recent posts, I’ve received a salvo of tiresome rationalizations. Here’s a typical one:
“On the other hand, it is not necessary to invoke psychometric alteration in order to suggest that GCSE and other testing regimes have been changed in order to (among other imperatives) minimise the apparent statistical gap between negroes and other ethnic groups. The way in which this has been achieved, it would appear to me, is that the tests have been sufficiently dumbed down and made stultifying, that they do little to discriminate amongst the right-hand side of the bell curve and therefore”
One way to evaluate whether the above is true is to look at the variance in scores and the g-loadings of the tests. Alternatively, we can compare the racial differences to class differences. The latter we know are, in part, due to genetically conditioned differences in intelligence. Below are graphs for indexes of cognition by race and class.
So in the LSYPE data, the the highest 1/5th, in SES, scores about 1 SD above the lowest 1/5th. Which is comparable to the difference in the US. The ethnic differences, however, aren’t.
Graph 1 — Goodman et al., 2009. Inequalities in educational outcomes among children aged 3 to 16.
Graph 2 — Goodman et al., 2011. Children’s educational attainment and the aspirations, attitudes and behaviours of parents and children through childhood in the UK
There is not a truth existing which I fear or wish unknown to the whole HBD community — Chuck
The data in your graph is from two different studies and cohorts and is taken at different times. Are you sure this data is representative? I think the LSYPE uses the actual GCSE that the child took, obtaining them from the school. But we know that all children in England do not take the GCSE. If lower IQ kids at 5 are less likely to take the GCSE that could easily explain the closing of the gap over time.
His point was that I cobbling longitudinal studies together. Others questioned my GCSE data. For example, Matt argued:
GCSE gaps overall seem pretty awful as a general measure of cognitive ability, despite being great for universities, because it is quite possible to take a set of GCSEs in “Art and Design”, “Performing Arts”, “Music”, “Psychology”, “Hospitality”, “Health and Tourism” and “PE”(!).The core GCSEs are English, Maths and Science, which we’d guess have a respectable IQ loading, but the others may be less certain. Those set of GCSEs which have the same content (mathematics and english) as the SAT are probably a good substitute for the SAT (although less than perfect, because, yes, coursework).
Matt’s point was that GCSEs scores, per se, can be misleading and that we should look specifically at Math and Reading test scores, which show relatively high g-loadings. Refer to other points made here. In response — and in followup with a point made by Galtonian — I offer a more recent analysis which circumvents these criticisms. Below are Math and Reading standardized test differences, over the years 1998 to 2007, based on a large cohort (N = 469,848). The scores are relative to a white mean set to zero, with standard deviations of 10:
This confirms my previous conclusion: The Black-White Gaps are trivial to small (in effect size) and don’t increase with age. The data here disconfirms Lynn’s hypothesis. Defenders of this embattled hypothesis need to account for the absence of more than trivial to small gaps, given the large differences between social classes and, more importantly, given the substantial correlation (.7) found between g and these test scores. The Chinese sample, of course, suggests that more than g is involved in population differences — but before you dismiss these achievement scores refer yourself to Gottfredson’s Implications of Cognitive Differences for Schooling Within Diverse Societies, especially table 3 and page 28 to 30. I’m sure that this paper is quite familiar to many of you. In that regards, at very least you have to grant that the near absence of an achievement gap in the UK is as much evidence against a UK IQ gap as the presence of an achievement gap in the US is evidence for a US IQ gap. Hereditarians have routinely argued that the US achievement gap is evidence of an IQ gaps (for example, Rushton and Jensen 2010, section 4), so my evidence should be in good standing.
Dustmann et al., 2011. Ethnicity and Second Generation Immigrants, Chapter 15 in The Labour Market in Winter: the state of working Britain 2010, edited by Paul Gregg and Jonathan Wadsworth, Oxford University Press, 2011
[Edit: Corrections made as needed. I guess I need to start looking into the nootropics.]
[Note: To see the evolution of my thoughts on this refer to my others posts on the UK gap: Occidentalist, 2011. The General Mental Ability (GMA) of Black British; Occidentalist, 2012. Is (global) “race realism” still tenable?; Occidentalist, 2012. Partially falsified. You will notice that I have been all over the place on this topic. The problem is that it’s difficult to get a hold of UK IQ data broken down by ethnicity and it’s even more difficult to interpret this data, given various confounding factors such as migrant status. I’ve contacted a number of psychologists working in the UK but no one seems to have — or is willing to share — information. (All of them were kind enough to reply, though.) Generally, seven points are noticeable concerning the Black-White gap: (1) Longitudinal studies show a small to nonexistent IQ gap between ages 3 and 11 (e.g., a Black-White gap ranging from 0-0.5 SD). (2) The Black-White Longitudinal IQ gap does not systematically increase with age (between 3 and 11) as it does in the US, but it either stays the same, decreases, or bounces around. (3) Longitudinal studies show a small to nonexistent achievement gap between ages 11 and 16 (e.g., A Black-White achievement test scores/GCSE point gap ranging from 0 to 0.35 SD). (4) Some cross sectional data shows a small achievement gap between ages 11 to 16. (5) Cross sectional data shows a moderate sized IQ gap by age 11 (e.g., a Black-White IQ gap of around 0.5 SD). (6) Some cross sectional shows a large adult gap, but there is little consistency across data — with points ranging from 0.3 SD to 1.7 SD — calling this data into question. (7) There has been a steep decline in the magnitude of the IQ gaps and achievement gaps over time.
My interpretation is: (1) There are, at most, small genetic differences, with respect to intelligence, between Black and White children born in the UK. I base this on the following: (a) There are currently trivial to moderate sized IQ gaps between children, ages 3-11, born in the UK as evidenced by longitudinal studies (e.g., MCS and EPPE 3-11) and as evidenced by achievement score gaps (e.g., KS2 Math and Reading test score differences). (b) The overall magnitude of the gaps are diminishing decade by decade across age cohorts, which suggests that some of the current gaps between children ages 3-11 have an environmental origin (for reference, compare the data in Lynn 2008 with Occidentalist 2012). (2) There are, at most, small genetic differences, with respect to intelligence, between Black and White adolescents and young adults born in the UK. I base this on the following: (c) The data from the mentioned longitudinal studies does not agree with the conjecture that the gaps increase with age. (d) This induction is confirmed by the finding, based on the LSYPE, that the gap in achievement scores, scores which are highly correlated with g within populations, do not increase with age. (e) Further confirmation comes from cross-sectional data, which shows trivial gaps between Blacks and Whites and yet large gaps, as expected by a genetic hypothesis, between children in line with the skill level/educational level of their parents. (3) There are, at most, small genetic differences, with respect to intelligence, between Black and White adults born, respectively, in Africa and in Europe. I base this on the following: (e) There are modest to large adult IQ gaps in the UK. Since these are not due to the low performance of Blacks born in the UK (points 1 and 2), it stands to reason that these are due to the low performance of recent African immigrants, who have low IQs due to being reared in low IQ affecting environments. This is evidence, then, that Black migrants to the UK are not super-duper genetically selected and unrepresentative, with respect to IQ, of their native populations. (f) As further evidence against super-duper selection theory, I present the result of Somalian refugees, who presumably were not very selected and who perform only 0.5 SD below Whites in measures of scholastic achievement.
As much as I disrelish the idea of Lynn et alia being wrong (about genetic racial differences), I can’t but help conclude that they mostly are (especially Nyborg with his ludicrous figures).
I sent out a number of inquiries for UK IQ data broken down by ethnicity but came up empty handed. Steve Strand, though, an accomplished educational psychologist who frequently conducts research for the Department of Education, forwarded me several of his recent papers on ethnic achievement gaps. For example, here, here, here, and here. It will be noticed that the standardized differences, presented in these papers, between ethnoracial groups (e.g., Blacks, Whites, South Asians, etc.) are small to trivial compared to the differences between classes.
When it comes to the gaps, it’s informative to look at the findings from longitudinal studies, as these are less confounded by migrant status and as these indicate how the gaps vary as a function of age. Disentangling race and migrate status is vital as what is of interest is the performance of ethnic/racial groups reared in roughly the same environment. And naturally, discerning the relation between the magnitude of the gaps and age is important given the relation between environmentality and age, within populations.
In the longitudinal studies, many of which can be analyzed online here, we see that, rather than increase with age, the gaps dissipate:
The first two points above shows the gaps based on the Millenium cohort study, which began in 2000-2001. (Notice the dopey new face of England shown on the MCS site.) The children were tested with the British Ability Scale. Missing is the Age 7 data, which I previously linked to. It will be noticed that between ages 5 and 7 the gaps narrow, not widen. The age 11 follow up for the MCS is being conducted this year and should resolve the issue of whether the IQ gaps narrow or widen with age. The Last three points show the results from the Longitudinal Study of Young People in England, which began in 2004. The cognitive measures reported are the Key Stage 2, 3 and 4 (GCSE) scores. You can see an alternative analysis of the data, broken down by test scores and point scores in tables 1 and 2 here. It should be noted that GCSE scores are derived from both achievement test data and other evaluations; this renders the relationship between population differences in GCSE and IQ scores ambiguous, even though IQ scores correlate highly with a latent GCSE factor within population. Using just the KS2 (age 11) data, which is based exclusively on math and reading test score results, the Black-White gap is a mere 0.3 SD. Other Longitudinal studies, however, such as EPPE 3-11, show, by age 11 (2008), no differences in Math and Reading ability tests. Refer to table Table A3.1: Cognitive attainments at the end of Year 5 by ethnic group, here.
Needless to say, the above is not what a moderate to strong genetic hypothesis (e.g., Lynn’s 50/50 global genetic/environmental hypothesis) would predict. In response to the above, defenders of hereditarian differences inevitably reply with the epicycle that immigrants are supper-duper selected. When I point to the performance of groups which are clearly not very selected such as Surinamese in Holland or Somalians in the UK, it’s more epicycles. It might be the case that there are some genetic differences, but based on the migrant data the magnitude of the difference between representative samples is likely small.
Generally, it’s not going to be the presence of racial differences that ushers in the twilight of the European peoples, as Lynn has argued, but the absence of them. Expect complete integration to move on pace.
Here are the age 7 g scores from wave 4 of the Millennium Cohort Study. The kids were tested in 2008 on the British Ability Scale. The first column shows the gQ, the second column shows the standard error, and the last shows the sample sizes. The Black-White differences is a mere .36 SD. For comparison, Flynn and Dickens (2005) gives an age 7 US difference of .5 SD.
See also my post: Partially falsified
Relatively large decreases in racial gaps over a decade don’t increase my confidence in a global genetic hypothesis. Below are the dutch CITO gaps (SDs = 10) as reported in Gijsberts, et al., 2012. The CITO is given by 85% of Dutch primary schools to age ~12 students; the sample sizes are large. Bartels et al. reports a 0.63 age 12 correlation between CITO and IQ and a narrow heritability of 0.60. Lynn (2006; 2008) considered the test to be a sufficiently good enough measure of g to include it in his discussions of race differences (see: Lynn, 2006; Lynn, 2008).
Ignore the White-Antillian gap. As Suzanne Model has shown in her exhaustive analysis of the performance of Black Caribbean emigrants to the US, UK, Canada, and the Netherlands, the parents of these kids are fairly selected. Black West Indian emigrants perform approximately the same in these diverse places, suggesting that they are equally selected. In the UK and the US, circa 2010, 2nd generation Black West Indian children/adolescents performed, on tests of mental ability, about a half of a standard deviation below 2nd and 3rd generation Whites. So we might not be surprised that, as of 2008, West Indian Blacks perform only 0.75 SD below Dutch Whites.
Focus instead on the Surinamese-Dutch gap. The gap from 1994 to 1997 is consistent with the 2nd generation IQ gap reported by te Nijenhuis et al. (2004), so this can be taken as a good estimate of the overall IQ gap. Over 2/5ths of Surinamese emigrated to the Netherlands. They can’t possibly be very selected. For reference, here are the rates of emigration by educational attainment as presented by Docquier and Marfouk (2005):
As can be seen by looking at Rate of Emigration by Education, approximately 41% of the emigrants were low skilled and 50% were high skilled. The difference amounts to approximately 0.25 SD deviations. If one compares the number of working age residents by education to working age emigrants to the OECD by education, assuming low, medium, and high skill represent -1 SD, +0 SD, and +1 SD standardized deviations, respectively, above the mean, one arrives at a similar standardized educational difference. Assuming that the correlation between education and general intelligence is the same in Suriname as it is in the West, approximately 0.6, the selection for general intelligence on the account of education could only be 0.15 SD (0.25 X 0.6). Since there would be regression to the mean, the second generation could only be selected approximately 0.1 SD on the account of their parents’ educational selectivity (0.15 x 0.6). Similar analyses can be done to show that on other accounts, Surinamese emigrants to the Netherlands are not particularly selected for g.
To the extent that a hereditarian hypothesis can be reconciled with the small and continually narrowing gap, it must in terms of population composition. According to the CIA World Fact Book, Surinamese are 37% Indian, 31% Creole (mixed white and black), 15% Javanese, 10% African, and 8% other. According to Lynn (2008 p. 205), this population should have a phenotypic IQ of 85; assuming that the narrow heritability of the various global IQ differences is 50%, as argued by Lynn (2006), the IQ gap should be, at minimal 0.5 SD. The 2008 gap, then, is at the minimal level consistent with Lynn’s genetic hypothesis.
Gijsberts, et al., 2012. Bijlagen Jaarrapport integratie 2011
Lynn, 2006. Differences in Intelligence An Evolutionary Analysis
Lynn, 2008. The Global Bell Curve: Race, IQ, and Inequality Worldwide
te Nijenhuis et al., 2004. Are Cognitive Differences Between Immigrant and Majority Groups Diminishing?
Bartels, et al., 2002. Heritability of Educational Achievement in 12-year-olds and the Overlap with Cognitive Ability