Fortunately for you, if you don't want to use race to distinguish you can instead use these characteristics to define your subgroups:

1) kids with two married parents at home

2) kids of parents with a college or professional degree

3) kids of parents who work white collar salaried jobs

4) kids of parents who are native English speakers

If you check all those boxes you're going to see that the resulting subgroup does very well on all standardized tests, including PISA. It will also be predominantly white/European-American & Asian/Asian-American. Those four attributes make the biggest difference in educational attainment, and brown & black families are the ones who are being left behind.

For the next thought exercise, consider what can be done to address this shortcoming in our political, social & education systems.

Isn't this just delineating by wealth? i.e. race and/or those 4 other categories are just correlations.

I'd be willing to bet most of these still hold once you control for wealth.

> First, where is the evidence for all of that?

Sorry, what do you mean by “evidence”? I clearly said that I am talking about PISA results, and described what these look like. You can look these up yourself if you are curious about more details.

> Second, other than racism, I don't see why we would distinguish between the 'race' of the Americans.

For better or worse, this opinion is not shared by the mainstream American culture and policy-making circles. We take great efforts to distinguish between these groups in the context of educational outcomes. It is not surprising, because these different groups very much exist in very objective sense, and quite objectively have different educational outcomes, which makes it useful to distinguish between them for the purposes like the original question, which was whether the educational quality went down recently. My original point that in order to answer this question, you have to distinguish between these groups so that you don’t fall victim to composition fallacy, is largely orthogonal to any discussion about the causes of these disparities, because it still stands regardless of whether the causes are 100% biological or 100% cultural or 100% result of systemic discrimination or whatever.

> Unless you make the racist claim that math skills biologically depend on race, there is no reason to compare some Americans with people in Africa because they have similar skin color - it's absurd.

You are creating a really weird straw man, because I don’t think that even extreme KKK-style racists have much of an issue with skin color per se. Who are you arguing against here? I am extremely confused.

> I don't believe that you are unaware of the hate and discrimination against immigrants and minorities, justified by these same and similar arguments.

Ah, here comes smears and innuendos that by open discussion of clear, objective facts that are relevant to policy making, I’m causing some kind of nebulous harm to some unnamed people via some proxies. Just stop it, I don’t care, and nobody cares anymore either.

> Unless you make the racist claim that math skills biologically depend on race, there is no reason to compare some Americans with people in Africa because they have similar skin color - it's absurd.

Math skills correlate with IQ, and average IQ varies by racial groups.

Calling facts you don't like racist doesn't erase them, and coming into discussions just to accuse everyone of heresy has gotten pretty old for most people.

These may be uncomfortable facts, but pretending they don't exist and brow beating people for discussing them keeps us from understanding the world more fully.

While IQ may vary by racial groups, overall math attainment not at the very top couple of percent has far more to do with socioeconomic circumstances & parental education than it does any inherent potential.

This is almost certainly an SES effect; the evidence for biological causation is weak and has gotten drastically weaker with each successive GWAS study. Either way: it's a race war point, and those are unwelcome on HN.

https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...

I think the opposite is true. There is large gap between educational outcomes of white and black Americans even if you match them on SES, and every successive GWAS increases the amount of variance in outcomes explained by variants, despite the input data only allowing for extremely crude metrics like years of completed education instead of actual test scores.

Literally the opposite thing, to what you just said, is true: GWAS studies have slashed heritability estimates (which beg the question of biological causation, but also establish a ceiling to it), by something like 80%. To say that this wasn't the outcome biological determinists expected from GWAS would be an understatement.

There is clearly a cohort effect, of students we identify as Black, with markedly lower educational success. Again: that's almost certainly an SES effect. Which shouldn't surprise anybody, given all the other SES effects that apply to that same cohort of people.

No, you are completely misunderstanding what GWASes show. The previous estimate of heritability of things like intelligence are entirely unaffected by GWAS results. That existing GWASes explain small percentage of variance is the problem with the GWASes, not with the preexisting, giant, extremely well replicated heritability literature.

Assume for the sake of argument that intelligence is 100% heritable, and is a result of additive heritability of thousands of variants. Pre-GWAS tools would in that case find something like 0.9 heritability (due to attenuation because of measurement error). Now imagine you run a GWAS that includes only 1% of SNPs of the entire genome, and your intelligence proxy is “has any university diploma”, instead of more accurate measures like eg IQ. In that scenario, you would never expect the GWASes to get even close to actual heritability.

This is where we are today: our genome coverage is relatively low (especially around rare variants), while the metrics we have are crude. Of course we are not even close to heritability estimates! If someone told you that low explained variance of GWASes is consequential for classical heritability estimates, they were either confused themselves, or deliberately tried to misinform you (extremely common problem in the field, with a long tradition from people like Lewontin and Gould back in the day, to people like Gusev today).

> Again: that's almost certainly an SES effect.

You just repeat it this without in any way substantiating it, despite me being very clear that the effect does not disappear even when you match by SES.

No, that's not all the recent studies do; they also attempt to separate inherited environment, and estimate direct vs. indirect heritability.

It's funny, I can point to comments from IQ-fixated people on this site from 10 years ago saying that GWAS was going to settle heritability stuff. All that's gone now, of course. Now the work is apparently fatally flawed.

Please, tell me which GWAS study is claiming to estimate entire heritabilty. I am not aware of any. This is because their methodology simply does not allow it. They by design are going to miss some heritability, and the methodology is not powerful enough to be able to tell us how much they are missing. If you don’t understand this, I think you are very confused about what these studies actually say.

> It's funny, I can point to comments from IQ-fixated people on this site from 10 years ago saying that GWAS was going to settle heritability stuff.

Please, do, and bring receipts! You should refresh your memory as to where the whole debate was 10 years ago. At the time, many people still hopelessly argued that intelligence have nothing to do with genes, and the “IQ-fixated” people replied that GWASes will conclusively show that they do. Now that they have, the goalpost has shifted, and people who argued that intelligence has nothing to do with genes are, similarly to you, arguing that it doesn’t have as much to do as “IQ-fixated” people have claimed. Of course, in 10 years, the goalpost will move again.

I haven't spaced on this, but I'm going to take a beat to generate a more rigorous reply. In the meantime: I think the evidence, both from twin studies and from genomics, for genetic determination of intelligence is very weak, has gotten weaker, and describes an effect too small to be relevant to the gaps we're discussing in this story; bringing racial IQ science into a discussion like this is arson.

I'll do my best to back those points up tomorrow.

It's a little weird that you called out Gusev, above; on this topic Gusev, though a geneticist himself, seems more like a popularizer of current research than someone going out on a limb with his own. But check his references: the scientists whose studies he cites seem to be saying the same thing he is.

I was waiting for the reply, but I realized that you might not be able to send one or edit your comment anymore, so here is one from me to give you opportunity to do so. I’m really curious what you found about GWASes or the state of debate around them 10 years ago.

Different "racial groups" have different average brain sizes. They vary exactly as you would think they do. Seems very biological to me.

How is this "race war"? No one is saying people should be punished or are inferior. We're talking about why schools perform the way they do in the US, not some "war".

Look at the antecedents to the 'dang comments I just linked you to.

It looks like he's mostly concerned with flame wars and slurs, which makes sense since they destroy conversations driven by curiosity and turn them into bitter fights. That's not at all what is going on here.

The fourth comment down, and this is sorted by date not relevance, is almost isomorphic to this thread. I believe you didn't intend to start a "race war" (that's Dan's term, not mine), but the big subtext of HN's norms is being responsible not just for your intentions but for the effect your comment has on the thread.

It is not at all established that there are meaningful, coherent IQ differences between "races", and the process of hashing that out on this thread will be an attractive (and tedious) nuisance that will attract awful comments --- for instance, like one suggesting "brain size" differences between races are explanatory.

So how do you stop this from happening?

The OP pointed out a highly informative and important point and the majority of the negative responses to it have been from people who want to bash the American education system.

They are using cries of racism and bringing in irrelevant topics like IQ intentionally so that they can shut down legitimate conversation because they want to continue to bash the American education system.

It's not a highly informative point; it's a deeply contentious one that revolves around a narrow(ing) and disputed finding that is drastically smaller than the educational gap we're discussing.