I have a few issues with this post. First, it's a mistake to assume IQ is a perfect measure of intelligence, especially over long time scales. I'm highly suspicious of psychometrics.
Second, I don't understand US race dynamics, but would anyone care to explain why this guy split the data by race, then disregarded all Black data points? Furthermore, it's well-documented that IQ scores are culture-specific. If you're from a different culture than whomever developed the test, you will score lower. This could also apply to the micro-cultures of two races even within the same country.
Finally, he says explicitly that the "sample size is very small", which makes any results drawn from them suspect.
Some countries establish preferential admissions policies to give certain groups benefits and to harm other groups. For example, the US in the early 20th century, and the USSR in the later 20th century, aggressively discriminated against Jews. A group that has different admissions standards will skew your data, so if black candidates can get to the same school with SATs hundreds of points lower, the rise of this bizarre phenomenon will distort the data.
> I'm highly suspicious of psychometrics.
> it's well-documented that IQ scores are culture-specific.
Psychometrics is not BS. There's a mountain of data showing properties like IQ (i.e. g) are real, useful measures, not culture specific, and are highly heritable. These are inconvenient truths to a lot of people with egalitarian ideals so there's mountains of FUD clouding the cold hard facts about IQ.
Let me clarify. I don't think psychometrics are BS; I think they are real measures of certain skills. However, the results of IQ tests have certain, significant limitations that people often don't appreciate. In practice, the tests are overused and the results are over-generalized. This article is a prime example of my criticisms.
The g that shows up when doing factor analysis on intelligence test results is not necessarily the same concept we all have of innate, immutable intelligence. There are good reasons to believe it's not the same thing at all -- g is mutable, somewhat heritable, and quite variable between different cultures and countries. It's a broad correlation; it doesn't correspond to any particular fundamental brain mechanism that would tell us useful things about biology.
IQ is a useful (if not especially real) measure, but less so than other measures, it is culture-specific, and it is hardly "highly heritable"; Devlin et al. showed a while back that its narrow-sense heritability is about 35% at best, and most studies of IQ's heritability grossly overestimate it by failing to control for a multitude of environmental effects. And psychometricians don't seem to get how factor analysis renders their beloved general intelligence measure a tautology, so it's understandable why someone would regard their work as BS.
[Edit to add some of the "cold hard facts about IQ" that IQ pushers like the parent comment don't tell you: IQ's heritability varies with socioeconomic status from about 80% to virtually zero, and also increases with age, which makes any claim about IQ having a high heritability across the board extremely suspect.
Note also that despite the name, heritability isn't a measure of 'how genetic' a trait like IQ is; IQ pushers often take advantage of that misunderstanding. Heritability is merely a rough estimate of the ratio of a trait's variance explainable by genetics to the trait's total variance in a population; as such heritability varies between groups, between environments, and between times. It is neither necessary nor sufficient to demonstrate that a trait is 'genetic'.]
Second, I don't understand US race dynamics, but would anyone care to explain why this guy split the data by race, then disregarded all Black data points? Furthermore, it's well-documented that IQ scores are culture-specific. If you're from a different culture than whomever developed the test, you will score lower. This could also apply to the micro-cultures of two races even within the same country.
Finally, he says explicitly that the "sample size is very small", which makes any results drawn from them suspect.
(Karma Nazi says, "no upmod for you!")