Home / HBD / Blacks and Whites with the Same IQ Still Differ a Lot

Blacks and Whites with the Same IQ Still Differ a Lot

melaninbrain

The cognitive profiles of whites and blacks are different. So even if the full-scale IQ is the same, blacks and whites do better or worse on different tests. This is based on one study by Craig Frisby, two studies by Arthur Jensen, one by Cecil Reynolds.

For the same IQ, whites will be generally better at problem-solving and spatial reasoning, while blacks will be better at memory, procedural tasks and rote learning.

The Frisby WAIS-IV Study

In 2015, Craig Frisby and Alexander Beaujean gave 1250 people the WAIS-IV IQ test, aged 16-90, and looked at the black-white gaps on the various subtests.

Friby and Beaujean (2015)

And so what we can see is that the black-white gaps are the largest on block design and picture completion. Blacks do relatively much better on cancellation, which is where you are told to quickly cross out certain shapes on a page.

What this means is that not only is there a difference in full-scale IQ between blacks and whites, but the cognitive profile is different. Daniel Seligman describes it thusly:

“On average, whites do better on all the subtests, but their margin of superiority varies considerably from one subtest to another. Or look at it this way: If you took a sample of black and white children, all of whom had scored around 100 on the WISC-R — that is, the black kids in the sample were above the black average — you would expect to find significant black-white differences on six of the thirteen subtests. The average black kid would do better on Arithmetic and Digit Span; the average white kid would do better on Comprehension, Block Design, Object Assembly, and Mazes”

The Jensen WISC-R Study

In 1987 Jensen took 86 blacks and 86 whites aged 9 and 10 and recorded their overall IQ scores, but also their subtest scores, and found that on some subtests blacks did relatively better, and on others whites did relatively better:

Black-White Gap by Subtest in Jensen 1987 Sample

Subtest Gap in SD
Object Assembly 0.7
Block Design 0.7
Similarities 0.68
Vocabulary 0.56
Information 0.49
Comprehension 0.47
Arithmetic 0.41
Picture Completion 0.36
Picture Arrangement 0.34
Digit Span 0.1
Coding -0.02
FSIQ 0.73

Gaps are in white Standard Deviations of 15.

Now the subtests all have smaller differences than the full-scale IQ; this is because the full-scale IQ is an aggregate of the subtests. Think of it like components of a football team – the best overall team doesn’t necessarily have the best offensive line, or the best cornerbacks, or the best quarterbacks – but it has the best combination of these things.

Normally the black-white full-scale IQ gap is closer to 1 standard deviation, but this just happened to be a slightly above average black sample Jensen was working with.

But the takeaway is the huge difference in subtest scores. Moreover, the two tests which blacks do massively better on are the digit-span test and coding. The digit-span test is remembering a string of numbers and repeating them back. Coding involves matching numbers to shapes on a grid. On more abstract problems, such as assembling an object out of pieces, or designing a shape out of different colored blocks, whites tend to do better.

To illustrate this further, we can, using this data, make projections of what the black-white gaps would be on each subtest if the blacks and whites had equal overall IQ scores for different IQs:

Black-White Subtest Gaps Assuming Identical Full-Scale IQs at Different IQs and Linear Scaling of the Gaps

Subtest Gap in SD Gap with same FSIQ of -94.5 -120 -130 -140
Object Assembly 0.7 0.265 0.33655 0.36464 0.392465
Block Design 0.7 0.265 0.33655 0.36464 0.392465
Similarities 0.68 0.245 0.31115 0.33712 0.362845
Vocabulary 0.56 0.125 0.15875 0.172 0.185125
Information 0.49 0.055 0.06985 0.07568 0.081455
Comprehension 0.47 0.035 0.04445 0.04816 0.051835
Arithmetic 0.41 -0.025 -0.03175 -0.0344 -0.037025
Picture Completion 0.36 -0.075 -0.09525 -0.1032 -0.111075
Picture Arrangement 0.34 -0.095 -0.12065 -0.13072 -0.140695
Digit Span 0.1 -0.335 -0.42545 -0.46096 -0.496135
Coding -0.02 -0.437 -0.55499 -0.601312 -0.647197

If this is confusing, I made a bar graph showing how much more whites scored than blacks on various subtests when their overall IQ is the same. Negative numbers means blacks are scoring higher:

Black-White Difference by Subtest when Full-Scale IQ is the Same Assuming Linear Scaling (Based on Jensen 1987)

colormc

To clarify this further, we can look at this in terms of what the black IQ is relative to the white IQ (always defined as 100) at given percentiles based on a black standard deviation of 12.5:

Black IQ by percentile, and predicted “subtest IQ”

Percentile Coding FSIQ Obj / Block
50 (0 SD) ~91 86 ~82.5
84.1 (1 SD) ~105.5 98.5 ~94.5
97.7 (2 SD) ~118.75 111 ~106.5
99.9 (3 SD) ~132 123.5 ~118

And so we can see that, for object assembly and block design, the black score on that is only 6.5 points higher than the white average at the 97.7th percentile. In addition, IQ deviation from racial means systematically regresses toward the mean, and so for the second generation of high IQ blacks, these are the best predictions of their offspring IQ:

Black IQ by percentile, and predicted “subtest IQ” of the second generation of parents with IQs in the above chart

Percentile Coding FSIQ Obj / Block
50 (0 SD) ~91 86 ~82.5
84.1 (1 SD) ~98.25 92.25 ~88.5
97.7 (2 SD) ~104.88 98.5 ~94.5
99.9 (3 SD) ~111.5 104.75 ~100.25

This model probably underpredicts the number of blacks with extremely high IQs because those are likely due to epistatic effects. That said, the proportion of blacks who – according to this – do as well as the white average on the object assembly / block design tests for more than one generation at a time will be vanishingly small.

I put ~’s next to the numbers, because these are not deep analyses, just rough estimates based on the NLSY data showing a regression effect of about .5 for both whites and blacks that deviate from their racial means.

Kaufman-ABC Results

Jensen also looked at the results from the Kaufman ABC test using the same sample of blacks and whites, and these were the results by subtest:

White-Black Gap in SD on Kaufman-ABC

Subtest Gap in SD
Arithmetic 0.64
Riddles 0.64
Triangles 0.56
Spatial Memory 0.44
Reading/Understanding 0.37
Matrix Analogies 0.32
Faces and Places 0.26
Reading/Decoding 0.24
Photo Series 0.19
Hand Movements 0.18
Gestalt Closure 0.14
Word Order 0.1
Number Recall 0.04

The main difference that pops out is that whites had a huge advantage on arithmetic in this compared to the WISC-R results, even though these were literally the same people as took the above test. So the arithmetic is probably different on K-ABC than on the WISC. However, overall, you see whites doing the best on riddles, triangles and spatial memory, while blacks doing relatively the best on number recall, word order and photo series.

Comparing Blacks and Whites with Actual Same Full-Scale IQs by Subtest

Cecil Reynolds in 1983 actually found 270 blacks and 270 whites with a median IQ of 89, and so instead of having to statistically infer the subtest gaps for theoretically equal full-scale IQ blacks and whites, he just compared groups of blacks and whites that actually were equal, and the results were slightly different from the above projections:

Black-white gap by subtest of blacks and whites who all have a full-scale IQ of around 89

Reynolds (FSIQ matched) Gap in SD
Object Assembly 0.51
Mazes 0.5
Block Design 0.27
Similarities -0.11
Vocabulary 0.06
Information -0.16
Comprehension 0.44
Arithmetic -0.36
Picture Completion 0.11
Picture Arrangement 0.34
Digit Span -0.62
Coding -0.49

The results don’t necessarily add up to zero because of weighting.

The results here are similar to Jensen 1987 at the extreme ends, with whites doing relatively the best at Object Assembly and Mazes, blacks doing the best at digit span and coding. The real changes though are arithmetic – where blacks do relatively much better, and comprehension, where whites do relatively better, than they did according to Jensen 1987. This could just be the a result of real changes in IQ not scaling linearly with each subtest in a given direction.

That said, you still see whites doing better in more abstract tasks, with blacks doing better in more mechanical, concrete tasks when IQ is matched.

Now one criticism would be that these studies are old. But so is the black-white IQ gap, and the reality is that the Flynn Effect (the general rise in IQ scores) stopped in western countries roughly around the time these tests were being done. So while the data is old, the situation in terms of IQ and “g” hasn’t changed from when these tests were done.

But when you step back for a moment, think about the proposal that for whites and blacks must have precisely identical cognitive profiles when IQ is matched. Why would we assume this? Why would we assume that all of the different cognitive skills would just so happen to line up with populations that have spent tens of thousands of years evolving apart, when they don’t even have the same brain size and shape or the same overall IQ scores?

The point being that the whole structure of intelligence between blacks and whites is significantly different, full-scale IQ is only part of the story, and blacks and whites with the same overall IQ will have very different cognitive abilities.

  • Augen

    Blacks , whites have inate differences. . . . males, females have inate differences. . . . So?

    • Deplorable KEK

      So build the Wall deport them All!

      • Nemon

        Top Kek!