“People so much wanted the story to be true . . . that they couldn’t look past it to more mundane explanations.” – Angela Saini, Superior.
I happened to be reading this book around the same time as I attended the Metascience conference, which was motivated by the realization during the past decade or so of the presence of low-quality research and low-quality statistical methods underlying some subfields of the human sciences.
I like Saini’s book a lot. In some sense it seems too easy, as she points at one ridiculous racist after another, but a key point here is that, over the years, prominent people who should know better have been suckers for junk science offering clean stories to support social prejudices. From Theodore Roosevelt in the early 20th century to David Brooks and the Freakonomics team a hundred years later, politicians, pundits, and scientists have lapped up just-so stories of racial and gender essentialism, without being too picky about the strength of the scientific evidence being offered.
Superior even tells some of the story of Satoshi Kanazawa, but focusing on his efforts regarding racial essentialism rather than his gender essentialist work that we’ve discussed on this blog.
As Saini discusses, race is an available explanation for economic and social inequality. We discussed this a few years ago in response to a book by science journalist Nicholas Wade.
As Saini points out (and as I wrote in the context of my review of Wade’s book), the fact that many racist claims of the past and present have been foolish and scientifically flawed, does not mean that other racist scientific claims are necessarily false (or that they’re true). The fact that Satoshi Kanazawa misuses statistics has no bearing on underlying reality; rather, the uncritical reaction to Kanazawa’s work in many quarters just reveals how receptive many people are to crude essentialist arguments.
A couple weeks ago some people asked why I sometimes talk about racism here—what does it have to do with “statistical modeling, causal inference, and social science”? I replied that racism is a sort of pseudoscientific or adjacent-to-scientific thinking that comes up a lot in popular culture and also in intellectual circles, and also of course it’s related to powerful political movements. So it’s worth thinking about, just as it’s worth thinking about various other frameworks that people use to understand the world. You might ask why I don’t write about religion so much; maybe that’s because, in the modern context, religious discourse is pretty much separate from scientific discourse so it’s not so relevant to our usual themes on this blog. When we talk about religion here it’s mostly from a sociology or political-science perspective (for example here) without really addressing the content of the beliefs or the evidence offered in their support.
Tomorrow’s post: Laplace Calling
Andrew: Any thoughts on Steve Sailer’s review of Superior? In particular, here is Sailer quoting Saini:
Do you think that Reich is a “ridiculous racist?” If not, then how would you distinguish his views from others that Saini attacks?
Steve has written on Saini a number of times, e.g.
http://www.unz.com/isteve/a-not-so-scientific-racist-declares-her-ancestral-homeland-the-smartest-country-on-the-planet/
http://www.unz.com/isteve/science-angela-sainis-superior-is-not-politically-correct-enough/
Greg Cochran has piled on as well:
https://westhunt.wordpress.com/2019/07/28/doubling-down/
My impression from what I’ve read is that Saini is unable or unwilling to comprehend the current state of scientific understanding and entirely captive to self-motivated reasoning.
There are definitely some valid critiques of the book (as is the case for all books), but I disagree with the following statement:
“My impression from what I’ve read is that Saini is unable or unwilling to comprehend the current state of scientific understanding and entirely captive to self-motivated reasoning.”
I dont think the current state of scientific understanding regarding the causal effects of race is all that comprehensive which would explain why Saini or anyone cannot comprehend it.
A lot of research regarding race as a “determinant” or “cause” has been conducted by researchers who don’t use directed acyclic graphs, which are PARTICULARLY important here.
https://www.ijcai.org/proceedings/2019/0199.pdf
Directed acyclic graphs aren’t particularly important here or anywhere. They are a method for solving unrealistic toy problems and have never helped solve real problems.
I encourage readers of Dr. Gelman to read Ms. Saini’s book because, while you may well not believe me now, if you read her book you are more likely to come to agree with me that:
A. She isn’t very bright
B. She hasn’t thought very hard about her topic
C. But it’s still easy these days to get a lot of praise like she has received for just repeating the usual talking points.
Is it true Saini vowed to write this book (or do something similar) when she was 10 because some kids threw rocks at her? This makes it sound she made up her mind on these issues at the age of 10.
I found an interview with Saini that talks about the rock incident. The article makes clear the incident was an important motivator in her life, that it is why she became a journalist, but it does not say she made up her mind on scientific issues at that time.
https://www.npr.org/sections/codeswitch/2019/07/10/416496218/is-race-science-making-a-comeback
As I wrote in my review of Saini’s book:
Saini also goes to interview David Reich, who runs the world’s most productive lab for sequencing ancient DNA. (Reich’s 2018 book “Who We Are and How We Got Here” is so informative that I wrote three reviews of it: Intro, China vs. India, and Africa.)
Initially, Saini is wowed by Reich’s prestige and his polished lines about the ubiquity of ancient “migrations” (a euphemism for invasions, conquests, enslavements, epidemics, and slaughters):
“What we think of as “indigenous” Europeans are, Reich and other scientists now understand, the product of a number of migrations over the past fifteen thousand years, including from what is now called the Middle East.”
Saini takes everything personally in that girly style that predominates in 2019, gushing:
“This is the book I have wanted to write since I was a child, and I have poured my soul into it.”
As a loyal Indian racialist (her first book was “Geek Nation: How Indian Science Is Taking Over the World”), she has an obsession with finding sciencey-sounding arguments that her people have just as much right to move to England as the native English have to live there. Hence, she is much cheered by Reich’s finding that the Battle-Axe Culture steppe invaders exterminated most of the population of England about 4,500 years ago:
“When considered from the perspective of the deep past, race, nationality, and ethnicity are not what we imagine them to be. They are ephemeral, real only to the extent that we have made them feel real by living in the cultures we do, with the politics we have.”
Saini reflects:
“If skin color and genetic purity can’t be a measure of ethnic identity, because Britons have changed on both these counts over the millennia, then there’s nothing to prevent anyone from anywhere from earning citizenship and becoming truly British….”
Well, there is the law. But the law didn’t stop the Proto-Indo-Europeans (i.e., Aryans) from conquering Western Europe 4,500 years ago, so why should it stop anybody from doing it again today?
https://www.takimag.com/article/arguing-against-reality/
Steve:
Please leave the sexist comments at home when commenting here. If you have a disagreement, that’s fine, express it directly rather than through sexism.
Saini said:
“This is the book I have wanted to write since I was a child, and I have poured my soul into it.”
So it appears she did form her opinions when she was 10 and spent subsequent years seeking out and collecting the scientific and polemical bits needed to support those opinions. She must be very proud of how prescient she was … all the science lined up so nicely with her 10-year-old worldview.
Terry:
When Saini writes, “This is the book I have wanted to write since I was a child,” I see no reason to take this to imply that the specific content or perspective of the book was determined when she was a child. Lots of children are interested in ancestry, and some of those children want to write books, and some of those children have experienced racism. In that overlap there are many possibilities. I don’t see your supposition (“So it appears” etc.) of her book-writing method to be at all supported by the actual book at hand.
Andrew:
For many reasons, I stand by my interpretation.
First, my interpretation is straightforward. I believe her when she says “This is the book I have wanted to write since I was a child” “This book” is, roughly speaking, “a stunning indictment of scientific racism” (my words) and I believe she has wanted to write this book since she was a child. (We can quibble about whether she intended to only write a stunning indictment of racism or of scientific racism. Whatever.)
Second, this dovetails with her personal experience. In another interview, she talks about how racist kids threw rocks at her when she was 10. This was clearly important to her, and I completely understand why this made her passionate about combating racism.
Third, her statement comes from the intro to her book. The rest of the intro is chock a block with standard Social Justice Warrior tropes repeated uncritically. There is no nuance and no indication she has ever seriously thought otherwise. I believe her when she makes clear her passionate commitment to her ideas. There are many people like this … indeed, a primary criticism of hers is that racist scientists’ results are shaped by their racism. I see no reason to believe her findings are any less influenced by her deeply held convictions.
Fourth, she gives no indication of a “voyage of discovery” to find what to write about. There is no reason to believe she ever meant to write about anything else or seriously contemplated reaching any other conclusion.
Fifth, she calls herself a journalist. Many modern journalists see their craft as putting forward a pre-determined narrative in as forceful a fashion using whatever support comes to hand and ignoring unhelpful evidence. The editor of the NYT came pretty close to saying exactly this. I see no reason to believe she is not similar in this way to many other journalists today.
Sixth, I can’t speak to a lot of the science, but I can speak to the nuttiness of the argument that race isn’t real because it is a social construct (and closely related arguments). Race, colors, and dog breeds are all social constructs, but I don’t see how that means they aren’t real. This suggests she has little interest in being rigorous and will shape whatever is at hand to reach a pre-determined conclusion.
Sixth, her book seems to be a grab bag of rather old ideas. I’m not really sure why this supports my point. Perhaps it is relevant because it undermines the notion that her goal is to educate or report. This is all old news.
Bottom line: I believe her when she says this is the book she wanted to write since she was a child.
Footnote: she explicitly disclaims interest in her ancestry. She says her parents are “the only ancestors I need to know”. So that doesn’t seem to be the motive for the book.
Correction: The Saini statements are not in the Forward of the book, they are in the Afterward and Acknowledgements.
If you read Saini’s book carefully, you’ll notice that many of the strong emotions that affect her difficulty dealing with contemporary science seem to derive from a conflict between her parents and her grandparents. For example, the dedication of her book reads:
“For my parents, the only ancestors I need to know.”
She hints that her parents are from different Indian religions, perhaps Hindu and Sikh, and that this caused trouble with her grandparents.
As we see often these days, these kind of subcontinental communal prejudices wind up getting transmuted intellectually into reasons for blaming Galton for her unhappiness.
D:
1. I don’t think Saini is “attacking” anyone. She’s reporting.
2. I don’t recall Saini describing Reich as “ridiculous.” When I wrote above that Saini “points at one ridiculous racist after another,” I didn’t mean that everyone she wrote about was ridiculous or a racist, or both. Saini wrote about serious scientists studying race and biology, and also about some people, scientists and others, for whom racism seemed like a core belief. Beyond this, I accept that “ridiculous” is in the eye of the beholder, so if I describe some of the characters in her book as being ridiculous, that’s just my impression.
3. I followed your link and I don’t Sailer is describing Saini’s book accurately. For example, he writes that Saini casts Albert Einstein as a “villain.” That’s ridiculous. Saini quotes something racist that Einstein wrote. The point was not that Einstein was a villain; the point was to illustrate the prevalence of racist attitudes, even among scientists.
I would prefer to read it before I draw any inferences from Andrew’s post. As I may have mentioned on Andrew’s blog, prejudices are multicausal. I am interested, by virtue of being in Erik Erikson’s circles, child development prof. in the psychology and economics of ethnocentrism and racism. Influencers shape attitudes. I think some racists are increasingly being made to look ‘ridiculous’. justifiably. My opinion. In fact, some who I 1st met in DC who made racist comments would now not dare to do so b/c one of their leaders has signaled a shift in attitudes. That occurred similarly on the subject of gays, with greater acceptance. Some parents were surprised to learn that their offspring was gay.
Saini is not just reporting:
From a review of Superior in Nature:
====
Reich: “There are real ancestry differences across populations that
correlate to the social constructions we have.”
He adds: “We have to deal with that.”
But, as Saini notes, when racism is embedded
in society’s core structures, such research is born of the same
social relations
I’ve read Saini’s comment many times, and can’t figure out what she means.
If I study genetic differences between self-described Han Chinese and self-described
Yoruba Nigerians is that somehow invalid? What exactly is Saini criticizing here?
and does she (or anyone else) think what Reich said is false?
====
~
Nick:
Saini says that when racism is embedded in society’s core structures, such research is born of the same social relations. I don’t see why you take this as a statement that studying genetic differences between groups is invalid. I take it as a statement that we should be aware of the social context of such studies, not that the studies are, in general, invalid.
OK, then you think that Saini has no problem with Reich and I looking at
differences between Han and Yoruba? I sure didn’t read her book that way.
Nick:
I’m not saying that Saini has no problem with Reich’s research. She doesn’t seem so happy that he’s doing it. But I don’t think she thinks such a study is invalid, just maybe that for historical and political reasons, it’s a bad idea. That’s her view; I don’t think it detracts from her reporting that she has a position.
> I don’t think it detracts from her reporting that she has a position.
Really? Is there any position a reporter writing on topic X might have that would detract, in your view, from their reporting? Would a reporter who believed that vaccines cause autism be just as good as a reporter who didn’t, when they are both writing about measles? I doubt you really believe that!
My view would be that a reporter with a strongly held and widely publicized view on X is much less believable than a similarly talented reporter without such a strong view. That doesn’t mean that the reporter is always wrong or should never be trusted, but it does suggest one should be wary, as I bet you are when reading the work of reporters whose views are less consistent with your own views. (I am certainly guilty of this as well.)
I bet this explains why you/Nick read Saini so differently.
D:
Huh? Some of the best nonfiction books ever are by George Orwell. He had a position. From another category, another set of classic nonfiction books are by Bill James. He had a position. Jane Jacobs had a position. Etc. People often write about a topic because they care about it. Nick seemed to think that Saini was saying that a study of genetic differences between human populations would be invalid. I saw nowhere that Saini said or implied these things, nor do I think it relevant to bring up vaccine truthers, or whatever they’re called, in this discussion.
Is Saini in the class of George Orwell or Bill James? I don’t know. But I do think that reporters can and do write about things they care about. Indeed, the strength of some great reporters comes from their commitment to their views, coupled with uncompromising honesty about what they see.
It’s worth noting that at one point in the book, Saini quotes, approvingly, a researcher who thinks that scientists wanting to study genetic differences in population should be “vetted” for political beliefs. She acknowledges that this might be “heavy-handed” but doesn’t exactly retreat in horror from it. She has some deeply illiberal tendencies which make me deeply nervous.
Merryweather:
Saini quotes anthropologist Jonathan Marks as saying that people who “cannot handle the results shouldn’t be studying it. . . . We don’t want racists working on human variation because that doesn’t work. So it’s not a question of ‘should this be studied?’; it’s a question of who should be studying it and how should they be vetted.”
Saini continues by summarizing Marks’s argument. Then she writes: “To others, this sounds heavy handed. For example, William Tucker, the psychologist whose painstaking work helped expose the Pioneer Fund and the scientists it has backed—even the kind he abhors. . . .”
She continues: “Fundamentally, though, the problem is not the science itself. . . . The problem is how these ideas are used and abused in the wider society . . . For those with a political ideology to sell, the science (such as it is) becomes a prop. The data itself doesn’t matter so much as how it can be spun. . . .”
I agree that Saini doesn’t “retreat in horror” from Marks’s argument, but I wouldn’t say she endorses it at all. It looks to me like she’s stepping back and saying that a focus on the details of the claims of Kanazawa-type researchers misses the social point of the role that their publications play in social discourse.
I, like you, strongly disagree with Marks’s view that racists should not be working on human variation. There are a lot of racists out there; some are scientists; and it’s no surprise that some of them want to study human variation from a racist perspective. Rather than trying to suppress this, I think it’s better to shine light on it, which is what Saini does in her book.
Here’s what I wrote:
In Saini’s sprawling conspiracy theory about the malign forces that inspire evil scientists to keep on noticing differences between human groups despite seventy years of politically correct censorship, I am cast as a villain, along with, among others, polymath Francis Galton, psychometrician Arthur Jensen, geneticist James D. Watson, rock singer Morrissey, Harvard geneticist David Reich, and even Albert Einstein. (That lineup makes me feel like the batboy on the 1927 Yankees: honored just to be on the same field.)
Saini gets her story about me so wrong that’s it’s hard to have much confidence in the rest of her book.
https://www.takimag.com/article/arguing-against-reality/
Steve:
You write that Saini casts Albert Einstein as a “villain.” That’s ridiculous. Saini quotes something racist that Einstein wrote. The point was not that Einstein was a villain; the point was to illustrate the prevalence of racist attitudes, even among scientists.
You also say she gets her story about you “so wrong” but I don’t see what’s so wrong. You disagree with her on Jonathan Marks: she describes him as “genial and generous” and you describe him as having “the classic Angry Left Radical personality.” My guess is that Marks was genial and generous to her and angry to you. You have lots of arguments with Marks but that doesn’t mean that Saini got things so wrong, just that she told his side of the story rather than yours.
Also you write, “Saini reports that I, personally, facilitated the return to influence of nefarious ‘race science’ by starting a sinister email list in 1998. (Actually, it was in 1999.)” Here’s what Saini actually reported: “It was 1998 or thereabouts that an invitation arrived in Jonathan Marks’s email inbox. He can’t recall the precise date . . .” Later she writes: “By the summer of 1999, Sailer’s roster of members was astounding.” So, no, she did misrepresent the date: She accurately said “1998 or thereabouts” with a clear explanation of her source and her uncertainty.
Also, I didn’t see that Saini ever described your list as “sinister.” That’s your word, not hers.
Dear Andrew:
I’m guessing you don’t know Dr. Jonathan Marks. I don’t think he would like being called “genial.” He is famously non-genial.
We got along quite well. I admire him for his directness. Indeed, in 1998, Dr. Marks and I agreed to approach magazine editors to see if they would be interested in hosting us debating the emerging topic: “Does Race Exist?” Unfortunately, there was little interest in our proposed debate.
David Reich is a heavyweight. Angela Saini is a lightweight.
The funniest line in poor Ms. Saini’s book is after Reich sets her straight about the reality of race, and so she harrumphs:
“They are words I never expected to hear from a respected mainstream geneticist.”
I have not read the book. But I believe that Saini’s characterization of Reich is correct:
In your view, would this belief make Reich a racist?
D:
No, I would not say that this belief makes Reich a racist.
Here’s what I wrote in the context of Nicholas Wade’s book, which I did label as racist:
The term “racist” is not precise, as can be seen from the above definition: what is meant by “primary determinant” or “inherent superiority”? So I’m not claiming that “racism” is a sharp category. Racism is a way, or set of ways, of looking at the world.
For all practical purposes, Wade and Reich share the same beliefs. In particular, Reich believes that human populations (black and white, English and Irish, Icelanders and Japanese) have more than “more than superficial average differences,” including, presumably, on trains like “creativity, intelligence, tribalism.”
And, for the record, Wade does not believe that “race is the primary determinant of human traits.” He believes that genetics are very important and that genetic traits vary across population groups. And that is what Reich believes!
It’s one thing to say that race and genetics play an important role… it’s another entirely to say that “race is the primary determinant of human traits” as in the Merriam-Webster definition.
D:
Wade’s statements in his book were much stronger than the Reich quote above. For example, Wade talks about major differences developing in a few hundred years, while Reich is talking about minor differences after tens of thousands years. That said, I don’t know either Wade or Reich, and, while I read an entire book by Wade, I’ve never read anything by Reich, so I’ll make no comment on whether they share the same beliefs.
> Wade talks about major differences developing in a few hundred years,
Are you referring to Wade’s discussion of the Ashkenazim, primarily based on “Natural history of Ashkenazi intelligence“? If so, isn’t Wade right, unless you have some idiosyncratic definition of “major?” That is, in the 13 years since this article was published, there has been no complaints about its central claim of major genetic differences between Ashkenazim and other Europeans, differences which were most generated post 1000 AD. (There are criticisms as to whether/how this difference evolved. Some argue that it was neural drift and/or founder effects. But no one (?) doubts that genetic differences exist.
D:
From my Slate article:
Ah, yes. But Wade is fairly clear (isn’t he?) in labeling this as speculation. If I can summarize his argument:
1) We know from the Three Laws of Behavior Genetics that “All human behavioral traits are heritable,” which means party genetic.
2) We know that certain behaviors — tendency to save, tendency to commit violence — have changed over periods of hundreds of years, long enough for genetic change to play a role, although non-genetic change has happened as well.
3) Wade speculates that genetic changes explain part of the change. That is that the frequency of certain SNPs in England changed enough from 1000 to 1800 to matter.
He offers no proof of this, because there is none. But he also speculates that it might be true.
Do you think that it is impossible that this is true?
Assuming you answer “No,” what is your prior on it being true?
My own prior, for these specific cases, is 25%.
Or do you think these are ill-informed questions which can not be parsed in a sensible fashion?
D Kane: the more appropriate question is “how big of a role did genetics play in the observed behavior changes?”. No one cares if 0.02% of the change is down to genetics and yet it would be still true that it “explains part of the change”.
priors are basically pointless here because there is no measurement we could make which is informative since it’s completely unknown how genetics affects complex behaviors. It’s definitely the case that it does, but we can’t look at someone’s genome sequence and say “this person will like plaid shirts, striped pants, and socks with sandals”
No good scientist should pretend this science is settled. The divide (and Saini is clearly on one side) is between those who think one can study these things open-mindedly without being a racist, and those who don’t.
I don’t want to silence Saini. I don’t want to silence Kanazawa. (I don’t agree entirely with either of them.) I don’t want to silence Jensen, Wade, Reich, etc. I think Saini wants to silence everyone in a certain red zone, especially if by “silencing” you include social ostracism.
Why not just talk/argue it out?
Mike:
I don’t think Saini is silencing anyone. She’s doing reporting—it’s the opposite of silencing; she’s talking it out. She’s less interested in the question of whether some ethnic group is x% more innately criminal than some other group and more interested in the question of how these ideas get spread and how they interact with political movements. That’s fine—she can talk this out and you can talk out the aspects of this that you care about. And I don’t that Saini thinks that only racists can study these things open-mindedly.
The point is that open-minded discussion can go in many ways.
One kind of open-minded discussion is the sort that Steven Pinker likes to talk about: “Do African-American men have higher levels of testosterone, on average, than white men?”, “Is the average intelligence of Western nations declining because duller people are having more children than smarter people?”, “Is homosexuality the symptom of an infectious disease?”, “Would damage from terrorism be reduced if the police could torture suspects in special circumstances?” These are politically loaded questions which pretty directly connect to histories of discrimination and violence, and so quite naturally they bother Saini and others.
Another kind of open-minded discussion is more in Saini’s line: To what extent is modern race science a cover for racist beliefs? To what extent have these beliefs been pushed by ignorant rich people who’ve funded fringe scientists? To what extent is race science used as a justification for discrimination and violence? These are politically loaded questions can be seen as analogous to Stalinist suppression of open inquiry, and so quite naturally they bother you and others.
I think Pinker-type questions and Saini-type questions are fine. I think it’s good to talk about all these issues, while recognizing the very real historical resonances they have. I don’t think Pinker is anything close to a Nazi, and I don’t think Saini is anything close to a Stalinist.
P.S. I don’t agree with anything I’ve ever seen from Kanazawa. His theories (when taken qualitatively, not quantitatively) may have some validity, but his data analysis has essentially zero relevance regarding these theories, for reasons I’ve discussed many times. If he did pure theory, fine: I wouldn’t find it valuable, but it is what it is. Also if he did pure theory I don’t think his work would’ve received any of the positive attention that it got. His empirical work is terrible, and it’s the source of whatever positive reputation his work has. This doesn’t make all race science valueless; it’s just disturbing that anyone would take this work at all seriously, any more than they would take seriously the ESP work of Daryl Bem, etc.
P.S. Thanks for the comments. They help me think things through. Also, this blog would be pretty damn boring if everyone agreed with me all the time!
Nonsense. They are not the same. Wade makes the suggestion that the reason Thailand is poor and China is rich is because of the genes, and even if Thailand had the same resources and institutions, it would still be poor.
China is rich? According to what definition?
Wade and Reich share the same beliefs? Absolutely not.
Reich (and I) signed a letter criticizing Wade’s terrible book (A troublesome inheritance)
Google “Geneticists say popular book misrepresents research on human evolution” for the letter.
Wade clearly thinks that genetic differences underpin racial inequality.
Reich and I disagree, which is not to say that tens of thousands of years of separation will not induce
some average differences in some traits.
“creativity, intelligence, tribalism” are quotes from Wade not Reich. Tribalism is genetic?
Iron Age Britons were “tribal” and have genetics very similar to modern British.
Nick is of course absolutely right. There is an ocean of difference between Reich and Wade. I’ll add that Reich’s views are based on his deep, quantitative understanding of human population genetics, while Wade is really just repeating slogans and has no idea what he’s talking about.
I believe that this is the letter which Nick Patterson is referring to. Here are some quotes from Reich:
I interpret Reich (and Nick Patterson?) as asserting that it is possible that human populations will differ on some traits — like intelligence and creativity — in ways that are non-trivial. Is that interpretation correct? Once we establish what Reich/Patterson believe, we can discuss how/whether those beliefs differ from Wade’s.
I’ll reply just for myself.
1) It is certain that human populations will differ on average for some traits.
Nigerians have darker skins that Norwegians. Mbuti are shorter than Dinka. These two
examples are at least primarily genetic.
2) I am not a quantitative psychologist but suspect there are no reliable measures of “creativity”
3) IQ is a measurable phenotype with a complicated relationship to “intelligence” .
4) There are IQ differences on average with current populations, with as far as I know no good studies
disentangling genetic and environmental effects (such as racism) or cultural influence on test-taking.
Is it possible that there are genetic differences between groups in IQ. Yes, why would this be impossible?
Is there scientific evidence for this? — No.
And I object to the question being “Are Nick’s beliefs the same as Wade’s?” I signed a letter that I hope made
that clear.
> I signed a letter
With all due respect, that letter is not very specific with regard to your differences with Wade.
> There are IQ differences on average with current populations
You and Wade agree on this.
> no good studies disentangling genetic and environmental effects
Perhaps. But before diving into the literature, it would be useful to get a sense of what you would consider to be a good study. For example, consider a paper A combined analysis of genetically correlated traits identifies 187 loci and a role for neurogenesis and myelination in intelligence. This purports to identify specific genes associated with intelligence. Assume that this finding is true. If a future study established that SNPs at these locations vary, on average, across groups, would that be a “good” study?
Of course, any single study is contestable and can rarely prove anything on its own. I just want to get a sense of what sort of study you would characterize as good, even if it alone would not end the debate.
What about the belief that race is the primary determinant of human outcomes? It seems to be shared by both racists and so-called anti-racists.
Yyw:
Interesting point. I guess there are (at least) two ways of being an anti-racist:
1. Believing that different racial groups differ only in minor or superficial ways but not in major ways dealing with personality and intellectual abilities.
2. Believing that different racial groups differ in major ways, but opposing the political positions held by many racists.
The anti-racists described in Saini’s group pretty much fall into category 1, whereas you seem to be talking about category 2, which I think is a different story.
On personality, didn’t Harvard admission data show some interesting racial differences?
/s
If I remember correctly, David Card found the data to be good enough to include in his analysis.
Yyw:
My colleagues and I wrote about the Harvard case here. In any case, I don’t think it’s reasonable to make general inferences about racial groups from a small subset such as Harvard applicants.
Andrew,
I agree with you. I was just venting.
I would think most anti-racists belong to category 1, but it seems that many look to race and race-related mechanism first for explanation of outcomes.
> 2. Believing that different racial groups differ in major ways, but opposing the political positions held by many racists.
Would you say those in group 2 are “racial essentialists”?
“Believing that different racial groups….”
I’d say “believing” is the key word in this whole discussion.
Doesn’t look like there’s a lot of good science going on with this topic. It would be interesting to read Reich’s body of work. My expectation is that, if it’s even possible to establish that there is a measurable distinction between any two races in any physical, emotional or intellectual ability (we know there are some appearance variations) – I don’t think this has ever been convincingly established, right? – it’s almost certainly *not* possible to establish whether those differences are genetic or cultural (internal preferences etc) or cultural (external, racism etc).
This is just like picking trends out of hurricane data. You can find them. They’re just not meaningful for the purposes that people use them for.
I’m not sure about what do you mean by “measurable distinction between any two races in any physical ability”.
Would you say that there are measurable distinctions in any physical ability between people with sexual chromosomes XX and people with sexual chromosomes XY?
The confusion regarding race as a “determinant” of human outcomes is a function of science’s lack of understanding of causal inference. In the potential outcomes framework, race obviously cannot cause anything since we cannot intervene to change a person’s race. However, unequal race relations can cause outcomes and we can intervene to change unequal race relations. The only way to evaluate the causal effect of unequal race relations is to realize that this is a mediation problem. Despite the fact that the total causal effect of race is nonsensical, if you do a mediation analysis, you can still interpret the indirect causal effect of race as the portion of the racial disparity in an outcome that would be eliminated had a specific unequal race relation been eliminated.
Race–> Health Insurance–> Mortality.
Race doesnt cause death. Lower rates of health insurance among minorities compared to whites (unequal race relation) causes minorities to have higher mortality rates as compared to whites.
It is frustrating that many scientists who study how race “determines” outcomes have no formal training in causal inference.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4125322/
-Adan
Adan — “However, unequal race relations can cause outcomes and we can intervene to change unequal race relations. The only way to evaluate the causal effect of unequal race relations is to realize that this is a mediation problem.”
Well put. But this is so hard to explain to racist-adjacent laypeople. They think you are being an equivocating, politically correct social justice warrior avoiding the inconvenient truth about racial differences.
Kyle- Yes I agree with you. It can be challenging to explain this to laypeople especially when there is such variation in interpretation even among scientists. The trick is to make them realize that the framework I specified applies to any outcome, and any 2 races can be compared which means that unequal race relations can exist between any pair of races, in either direction.
In health, most of the time minorities experience worse outcomes compared to whites so we naturally look for an unequal race relation that favors whites over minorities. There are some outcomes where there is less evidence of any difference or evidence of very small differences that may be impractical to pursue to equalize. And every now and then, minorities will have better outcomes compared to whites ( for example in the ESRD population, minorities have better survival compared to whites). The above framework could be used to see what unequal race relation is contributing to better survival among minorities.
My understanding of the RAND health insurance experiment, and then the more recent Oregon ACA experiment, showed that giving people health insurance resulted in them consuming more health care but not having better health outcomes. Really in line with Robin Hanson’s theories. So in terms of U.S racial/ethnic groups there’s something called the “hispanic health paradox” because their health outcomes are better than one would expect given their socioeconomic status.
Wonks- Yes I should have chosen a better mediator because you;re right, there is still controversy regarding the causal effect of insurance. Perhaps a better example would be in cancer. Race–> stage at diagnosis–> mortality
And yes you are correct that there are some instances where minorities have better outcomes than whites (I also cited the ESRD population where minorities survive longer than whites).
However, when you look at all health outcomes, minorities consistently have worse outcomes as compared to whites (with of course some exception).
The causal inference framework that I cited could be used to compare any health outcome between races, no matter what direction the disparity goes.
However, when you look at all health outcomes, minorities consistently have worse outcomes as compared to whites (with of course some exception).
What data is this claim based on? My impression is that Asian-Americans generally have better health outcomes than white Americans. Hispanics also often have better outcomes than whites. It’s not about whites vs. minorities (an absurd dichotomy) but about the relatively poor health of some minorities, especially blacks.
New day so I will comment. Again, I have stated this above:
“The trick is to make them realize that the framework I specified applies to any outcome, and any 2 races can be compared which means that unequal race relations can exist between any pair of races, in either direction.”
I was just being general re: minorities vs. whites. It is generally accepted that large racial disparities in treatment, morbidity and mortality exist and in general whites seem to experience better outcomes. and minority group is certainly diverse.
https://www.ncbi.nlm.nih.gov/books/NBK220344/
https://www.ncbi.nlm.nih.gov/pubmed/29297054
https://bmchealthservres.biomedcentral.com/articles/10.1186/s12913-018-3180-0
https://www.cdc.gov/minorityhealth/chdir/2011/ExecutiveSummary.pdf
As Ive also stated above: there are many health outcomes that are very similar between races and those where there is a paradox. Its called the hispanic paradox because it goes against what you typically expect.
My main argument was that a causal inference approach is fundamental in order to understand how race
“influences/determines” outcomes as some have said. And yes, as a public health researcher, I can say that it is generally well accepted that racial minority subgroups experience worse health outcomes as compared to their white counterparts.
New day so I can comment agn:
Ive already stated that the causal inference approach I cited could be used to compare any health outcome between races, no matter what direction the disparity goes.
Youre right minority group is very different but in general Whites experience superior health outcomes:
https://bmchealthservres.biomedcentral.com/articles/10.1186/s12913-018-3180-0
https://www.ncbi.nlm.nih.gov/pubmed/26503034
https://www.cdc.gov/minorityhealth/chdir/2011/ExecutiveSummary.pdf
https://www.ncbi.nlm.nih.gov/books/NBK220344/
https://bmcpublichealth.biomedcentral.com/articles/10.1186/s12889-016-3744-z
There ar eplenty of outcomes that are similar between races and plenty where one minority group or more experiences better outcomes than whites. Hispanic paradox. Wouldnt be called a paradox if it was going against the norm.
Most public health professionals and clinicians generally accept the statement that whites experience better outcomes compared to whites and disparities in treatment and access to care are well documented.
The “hispanic health paradox” isn’t simply because hispanics are a minority, but because they tend to have lower SES, and among non-hispanic whites that would predict worse health outcomes than we see among hispanics. Asians are also a minority, but it’s not a “paradox” for them to have better health outcomes since they tend to have higher SES than hispanics.
any 2 races can be compared
https://bmchealthservres.biomedcentral.com/articles/10.1186/s12913-018-3180-0
https://www.ncbi.nlm.nih.gov/books/NBK220344/
https://www.ncbi.nlm.nih.gov/pubmed/26503034
most clinicians and public health folks accept that large racial disparities generaaly favor whites who experience better outcomes. with plenty of exceptions but disparities are well documented.
I don’t think that that hodge-podge of random articles allows one to reach the conclusion that “racial disparities generally favor whites who experience better outcomes.” And even if public health professionals generally believe that to be the case doesn’t mean it’s true. It’s easy to find data that contradict the conclusion. For example, this article on the largest health disparities between racial and ethnic groups in the US found (see Table 1) that among the health disparities identified, whites had the lowest prevalence for nine outcomes while Asians (often lumped together with Pacific Islanders) had the lowest prevalence for up to sixteen outcomes (differences not always significant given low sample sizes for Asians). The idea that white vs. non-white is a useful analytical dichotomy in social or medical statistics is unsupported by data and should be discontinued.
Most of the time race/ethnicity is categorized as Black White Hispanic Asian Other and not dichotomous. That being said race is even more complex than just 5 categories. But causal inference framework can compare any two groups.
I would survey physicians and ask them if they believe there are racial disparities because they actually see patients everyday.
Here it is Matt:
What you’ve been waiting for:
https://statmodeling.stat.columbia.edu/2019/07/04/41237/
I’m criticizing the idea that whites versus non-whites (or “people of color”) is a useful way of dividing the world. It isn’t. I agree that using multiple and more fine-grained categories is a much better approach and one that shows the fallacy of the white vs. non-white dichotomy.
Exactly. In fact, it was due to racially prejudiced assumptions like the above that of course White Privilege means whites live longer that almost nobody noticed the White Death, the growing mortality among poorer-educated whites for about 15 years until Angus Deaton and Ann Case finally managed to call attention to it a few years ago.
I doubt lower rates of health insurance cause death. Health insurance is a recent phenomenon, and the upper classes seem to be living just as long today as they always have (80s -90s; just look up how long historical figures lived who didn’t die due to violence).
It is more likely that lack of access to cheap energy/food/sanitation/safety causes earlier death. Burning oil and coal has probably increased life spans far more than health insurance.
There is some evidence that the ACA may not have been as impactful as we would have hoped. But thats just one type of insurance model. Innovative and different from previous insurance models.
However, in health policy literature, general forms of insurance such as fee for service Medicare have been shown to have strong causal effects on morbidity and mortality. If you disagree go speak to John Michael McWilliams:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2881446/
But with ACA causal effect of insruance so I agree it may not be best example.
That being said I 100% agree with you re: lack of access to cheap energy/food/sanitation/safety can also cause premature mortality. That means that they are also mediators of race and mortality disparities. The causal inference framework I cited allows for multiple mediators and SAS macros applying the potential outcomes framework can be easily applied to solve a multiple mediation problem.
I meant that access to modern healthcare has dubious net benefit at best, health insurance doubly so. A great example is Cuba with their chronic shortages and old equipment, yet similar life expectancy to the US.
You can also see that life expectancy in the US started dropping due to “poisonings” and suicides as soon as millions more people gained access to the US healthcare system starting in 2014.
Another thing to keep in mind is the effect of abortions on infant mortality and life expectancy stats. In the past those children would have been the most at risk of dying young, today they aren’t counted.
Infant mortality is about 5.8 per thousand births while abortions are 188 per thousand births. What happens if you start counting abortions as early deaths?
https://en.m.wikipedia.org/wiki/Abortion_statistics_in_the_United_Statesa
https://www.cdc.gov/reproductivehealth/maternalinfanthealth/infantmortality.htm
I would look into the work of John Michael McWilliams and Michael Chernew. ACA effects are controversial but modern healthcare has evolved from the ACA. We are now transitioning from volume to value based care and we are now barely starting to have enough follow up to evaluate recent experimental value based payment models including accountable care organizations, bundled payments and Medicare Advantage of course. Theres definitely evidence of cost reductions (the biggest problem in healthcare) and modest evidence for mortality:
https://www.ncbi.nlm.nih.gov/pubmed/30933606
https://jamanetwork.com/journals/jama/article-abstract/2727554
https://www.ncbi.nlm.nih.gov/pubmed/30183495
https://www.ncbi.nlm.nih.gov/pubmed/29541764
I got stopped at the first sentence of the first article:
This sounds like an arbitrary number to me. Why not also “adjust” for income, race, past healthcare expense, etc and all the possible interactions between these variables?
Ie, what got discussed here: https://statmodeling.stat.columbia.edu/2019/09/06/__trashed-2/#comment-1117789
Anoneuoid,
What do you mean by ‘poisonings’?
Both purposeful and accidental drug over/under-doses: https://www.bmj.com/company/newsroom/deaths-from-drug-overdose-and-suicide-outstrip-those-from-diabetes-in-us/
In a world with lots of racist cultural practices and policies, it’s entirely plausible that race could be a strong determiner of outcomes even without it having much direct effect on traits.
Are you referring to the US of today? I frankly don’t see how whatever racism there is today can have such a large effect.
The US government is obsessed with race. Just look at the space devoted to breaking things down by race in the demographic data they produce. It was disturbing to me…
Another good critique of Saini:
https://quillette.com/2019/06/05/superior-the-return-of-race-science-a-review/
“Saini wants us to ignore the basic tenets of Darwinism”
Interesting… My interpretation was: Saini wants us to ignore the basic tenets of SOCIAL Darwinism
That is a cogent intellectual critical review, but you are name-calling.
The critiques and I are both name-calling, except that they are calling something Darwinism and I am calling it Social Darwinism.
A technical comment from a geneticist.
Most genetic differences between human groups are not caused by natural selection
which is probably what is meant by “Darwinism” in this thread, but by genetic drift.
This is a purely statistical effect that must occur in populations not in genetic contact,
unless there are balancing (selective) forces acting to cancel the drift.
Most, perhaps. Although there are clear exceptions. Adaptations for oxygen at high altitudes, dark skin near the equator. I don’t think this general controversy is about that. But I’m not gonna get involved in this in the comments section of even a good blogger. If you don’t think the critiques that have been posted here call into question Saini’s competence, that’s all I need to know.
Thank you Nick. I am not as a geneticist.
As a geneticist can you help me to understand something: Why do folks equate genetics with race? Isn’t there more genetic variation within races than there is between races? Is automatically thinking about genetics when the topic of race is brought up troublesome?
I read this but I am not qualified to assess its credibility:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1893020/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1196372/
Adan:
Why do folks equate genetics with race?
That’s not a question about science but about popular perceptions.
I’m not really qualified to answer, and a coherent answer would take us far afield.
However there are for sure genetic differences between human groups. if that were not
true human population genetics (which is what I practice) would be impossible and
outfits like 23andme would not be in business. Incidentally most genetic differences caused
by drift (which is what I intensively study) have no effect at all on phenotype and therefore
are not under selection.
Yes! Based on my low IQ understanding of genetics, I run under the assumption that there are genetic differences between human groups.
I should have said “Why do some scientists equate genetics with race?”
My interpretation of the paper I linked to is that two individuals of the same race are more genetically diverse as compared to two individuals of different races.
This doesnt mean that there aren’t genetic differences between 2 racial groups. It just means that those differences are smaller as compared to the genetic differences observed among 2 individuals from same racial group.
To me, trying to understand how genetic differences between races causes differences in human outcomes will not be as fruitful since you will only explain the minority of the variation.
> My interpretation of the paper I linked to is that two individuals of the same race are more genetically diverse as compared to two individuals of different races.
Your interpretation is wrong.
The correct interpretation is: among all the things that vary from one person to another, things that vary consistently with race are a small fraction.
Hi Adan,
I see Mike has already replied with the study showing how self-reported race could be predicted almost perfectly by genetic analysis – already some 15 years ago.
To that I would add that the phrase “more genetic variation within races than there is between races” that you used is a common enough misunderstanding that it has acquired its own name: Lewontin’s Fallacy. See here for more: https://infoproc.blogspot.com/2008/11/human-genetic-variation-fst-and.html
Thanks pwyll. This is not my area of expertise but I am willing to try to learn.
What is the correct interpretation of this paper?
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1893020/
In particular, what does this phrase mean to you:
“The fact that, given enough genetic data, individuals can be correctly assigned to their populations of origin is compatible with the observation that most human genetic variation is found within populations, not between them”
it means just what it says…
suppose you had two factories from which cars came. If you look carefully enough at the cars you can find information about them that identifies the factory…. but it’s still the case that things that vary widely even from one car to another at the same factory: paint color, options packages, flawed parts, and assembly variation are far larger sources of variability than the fact that one factory consistently used bulbs from supplier A and the other factory consistently used bulbs from supplier B for example.
Lewontin’s fallacy and the “Within greater than between” comprises two related fallacies.
The first is that somehow this rules out important genetic differences between populations. It doesn’t. We can see with our own eyes racial characteristics that allow identification to genetic ancestry at extremely high levels of accuracy. These differences are “important” in the sense that they are obvious. There is no reason to declare a priori that traits important in other ways (like we value them) shouldn’t have genetic differences between populations.
The second is more technical, and it is that genetic differences may be correlated. This would be expected when natural selection causes differences between populations (and it would be shocking if populations separated for centuries who reside in notably different environments were not subject to differential selection).
One author of this article has a publishing history with Quillette that seems relevant: https://en.wikipedia.org/wiki/Noah_Carl
And this guy was an editor there right: https://www.reddit.com/r/Portland/comments/cw5x54/rightwing_star_andy_ngo_exits_quillette_after/ ?
From the article:
“Note that since the book is quite poorly structured, and in some places contradictory, it is not always easy to discern what Saini is or is not asserting. Nonetheless, we believe that the four propositions above comprise a fair summary of her main arguments.”
So they’re not sure what they read, so they’ve decided they’ll debunk the four propositions they came up with?
Andrew, you’re an expert in robust statistics, and have written repeatedly about some of the worst excesses of the replication crisis. So it seems strange to me that your blog posts like this one and the ones you’ve posted on Watson have been so stats-lite.
Both the study of intelligence and the study of human genetic variation are extremely stats-heavy… in fact as I recall many of the central concepts of statistics were originally developed for the problem of analyzing intelligence.
I’d be very interested to see you or one of your team members do a in-depth analysis of, for example, how statistics are used in defining IQ, and how your more advanced toolbox might be able to refine the concept, along with its interaction with genetics. (Steve Hsu has done the most accessible work on this that I’ve seen so far.)
I think you’re uniquely qualified to raise the level of dialogue higher than “these people are OK, but *those* people are racists”, and I’d love to see you tackle the problem on this blog. Is there analysis you’ve done that I’ve missed?
Pwyll:
I haven’t looked into this, but you could check out the work of Jelte Wicherts. He seems to know what he’s talking about.
Thanks. I thought that name sounded familiar but wasn’t sure where I had heard it before… doing some cursory searching it looks like he’s done some work debunking stereotype threat: https://emilkirkegaard.dk/en/?p=7534
…though I don’t immediately see anything on race or IQ.
I guess my gripe is that even though you’re an eminent statistician and this is a premier stats blog, posts like this one that are happy to condemn people as racists are curiously light on the core discipline of statistics… which makes me think that perhaps the reason they’re not being condemned on statistical grounds is that actually their results are pretty good.
I haven’t dug deeply into Kanazawa’s work – I trust you when you say that it’s shoddy. But when slurs are used here, instead of stats to condemn e.g. Watson, I start to get very suspicious.
Pwyll:
Opinions will differ, and I respect that. Different people have different experiences, attitudes, and perspectives.
I have no criticism of James Watson’s statistical work. As far as I know, he doesn’t have any—but, if he does, I haven’t seen it. I label Watson as a racist because he says lots of racist things, like “[The] historic curse of the Irish . . . is not alcohol, it’s not stupidity. . . it’s ignorance. . . . some anti-Semitism is justified. Just like some anti-Irish feeling is justified . . .” and “There is a biochemical link between exposure to sunlight and sexual urges.. that’s why you have Latin lovers” and “The one aspect of the Jewish brain that is not first class is that Jews are said to be bad in thinking in three dimensions… it is true.” Also “Indians in [my] experience [are] servile.. because of selection under the caste system” and “People who have to deal with black employees find [that they are equal] not true” and “East Asian students [tend] to be conformist, because of selection for conformity in ancient Chinese society.” etc etc etc.
Look at another way. Lots of people are racists. That’s just the way things are. Some of those racist people are going to be scientists, even prominent scientists. Racism is considered rude in modern society, hence the prominent scientists who loudly express racist views will get a lot of attention. James Watson, prominent scientist, expresses racist views all over the place and gets attention. Calling James Watson a racist is not a close call. Calling Watson a racist is like calling Bill Simmons a Boston sports fan or calling Philip K. Dick a paranoid or calling Evelyn Waugh a snob. These people wear it on their sleeve.
I appreciate your continuing to engage on this – thanks, and apologies that I haven’t had time to respond again until now.
I’d argue that all of Watson’s rude statements you highlighted are actually *are* “statistical work” in a sense… they’re statements that, to some degree, can be evaluated using statistical tools for how true they are.
Take his assertion about Jews being “bad in thinking in three dimensions”… what data exists on the subject? How does the distribution of Jewish 3D thinking ability actually compare with that of other ethnic groups? Is there only one general factor of 3D mental visualization, or are there several?
Or take the assertion about East Asian social conformity. Again, what does the data say? Would you have an opinion of the quality of the statistics in this paper, for example? https://www.researchgate.net/publication/262193686_Large-Scale_Psychological_Differences_Within_China_Explained_by_Rice_Versus_Wheat_Agriculture (One opinion on it: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4405999/ )
In other words, is Watson just rude, or is he rude and also wrong? And if he’s wrong, *in what ways* is he wrong?
“Racist” in contemporary America is a word with a lot of power to do harm… People such as Amy Harmon have done massive damage to Watson’s career and legacy by painting him as one. Saini also seems to be eager to paint people as racist, despite being somewhat of an “Indian Supremacist” herself. In fact, since many people argue that only white people can be racist, it seems to be morphing into, simply, an anti-white slur.
If I want to read about how someone is racist, I have many options for doing so. But what only a few people (such as yourself) can supply is a nuanced exploration of the science, and limits thereof, behind a given person’s controversial statements.
Pwyll:
You write of people “painting” Watson as a racist. Watson says things like, “some anti-Semitism is justified. Just like some anti-Irish feeling is justified.” Not lots of painting required here!
Again, think of this from a statistical perspective. Lots of people are racist, and some of those racist people are going to be scientists, even prominent scientists. Lots of history on that one. So when Watson says a lot of racist things, then, yeah, it seems about right to label him as such.
Then there Watson’s statements that potentially have empirical content, such as: “There is a biochemical link between exposure to sunlight and sexual urges.. that’s why you have Latin lovers” and “The one aspect of the Jewish brain that is not first class is that Jews are said to be bad in thinking in three dimensions… it is true.” Also “Indians in [my] experience [are] servile.. because of selection under the caste system” and “People who have to deal with black employees find [that they are equal] not true” and “East Asian students [tend] to be conformist, because of selection for conformity in ancient Chinese society.” Sure, some of these statements could be statistically true, some could be false. I think the idea that genetic selection is a ridiculous explanation why Watson views Indians as “servile.” It’s the sort of sentiment that traditionally would be heard after a few drinks at the country club. But, hey, who knows? I think Saini put it well when she wrote, “People so much wanted the story to be true . . . that they couldn’t look past it to more mundane explanations.” Country-club-style explanations of social and political inequality are comforting for the Watsons of the world.
One more thing: You write, “Saini also seems to be eager to paint people as racist.” I don’t see that at all. Saini in her book wrote about a lot of racists—no surprise given the book’s topic. She reports their words and their actions. I have no reason to think that Saini is “eager” to describe these people’s racists words and actions. And, whether or not it’s true that “many people argue that only white people can be racist,” this is certainly not an argument that Saini is making.
Andrew:
Were there any parts or aspects of the Saini book that you found troublesome in any way?
Terry:
No, not really, I don’t recall anything troubling about Saini’s book. I mean, sure, some of the stories were troubling—it was horrible to read about some of the things that people had done, and some of the racist science would’ve been funny except for the social influence it’s had, which makes it disturbing—but Saini’s treatment of all of it seemed reasonable to me. In this comment thread and elsewhere there have been claims that Saini got things wrong in various places, but when I tracked down the particulars, I didn’t see any problems with what she wrote. I am of course open to the possibility that Saini made mistakes or is missing big parts of the story, but I haven’t seen any evidence for that.
Hmmm, let me think for a moment. I do think there were one or two places where I disagreed with Saini, or where I felt she came on too strong. II can’t remember them offhand, though. I just picked the book off the shelf and took a look but couldn’t find these spots in a quick scan.
My final comment on this post is that a more rigorous approach to causal inference that uses directed acyclic graphs will help us understand the following statements better:
“Racial disparities in treatment and outcomes exists in medical populations”
“It is also compatible with our finding that, even when the most distinct populations are considered and hundreds of loci are used, individuals are frequently more similar to members of other populations than to members of their own population”
The last statement comes from:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1893020/
“It is also compatible with our finding that, even when the most distinct populations are considered and hundreds of loci are used, individuals are frequently more similar to members of other populations than to members of their own population”
No one knowledgeable has ever disputed this statement, including those who believe that genetic differences may be important causes between certain population differences.
I don’t think DAGs would help at all. To study racial disparities in treatment and outcomes in medical populations you need good research designs that can tease apart genetic and environmental causes of those disparities. Drawing diagrams doesn’t help because the perfect knowledge needed for such graphs is never available.
The “hundreds of loci” statement is easy to understand. When the resolution of the analysis is a few hundred loci, sometimes it won’t discriminately perfectly between geographically distinct populations. If you increase the resolution to thousands of loci, it does:
It breaks down, however, with data sets comprising thousands of loci genotyped in geographically distinct populations: In such cases, ω becomes zero.
An even more intriguing and perhaps novel power dynamics predicated on Pigmentocracy which has been alive and well in nonwhite majority societies since the end of the British empire.
Andrew,
Shouldn’t you be talking about the field of statistics in this post, whose foundations you and your colleagues teach every single day, and which were developed in pursuit of eugenic and racist agendas? Galton? Fisher? Pearson?
Who is Kanazawa compared to them?
Ed:
Yes, racism was central to some extremely important work in the history of statistics, and Saini does talk about Galton and his followers. I didn’t see the Galton story as central to Saini’s book, but, sure, his ideas are an important historical example of scientific racism. Galton made some mistakes but, yes, we teach his statistical ideas every day.
I don’t see much of a parallel between Kanazawa, on one hand, and Galton, Fisher, etc., on the other. Kanazawa is a sociologist who’s received some media exposure from time to time but has not had much scientific or cultural influence, he’s used statistical methods in a sloppy way characteristic of the recent replication crisis. Galton, Fisher, etc., have been extremely influential, inventing entire new fields in statistics and biology. Racism is pretty much the only thing that Kanazawa has in common with Galton, Fisher, etc.
That is kind of my point. When you said:
“A couple weeks ago some people asked why I sometimes talk about racism here—what does it have to do with “statistical modeling, causal inference, and social science”?”
And apparently you didn’t even think to mention some of the founders of statistics, I was a bit surprised.
Ahhh, yes, good point! I guess one reason I didn’t think of this is that, to me, I can separate their statistical methods from their racism. Even though, for Galton and many of his successors, the statistics, the biology, and the racism were intertwined.
And I should also say that the historical foundations of my field, anthropology, are also shot through with racist and eugenic agendas.
View on the context of these discussions from outside of the US, based on a recent visit to the US as a mixed race family: “A melting pot” that has never been heated up. Many groups living politely alongside together in the same society.
“The Flaw of Average” – just sayin’
Andrew,
This quote of yours sums up, to a great extent, my own perspective. I characterize such explanations as ‘incomplete theorization’ for they are attempts at theorizing.
‘My feeling about Wade’s genetic explanations for economic outcomes is similar to my feeling about other all-encompassing super theories: I respect the effort to push such theories as far as they can go, but I find them generally less convincing as they move farther from their home base. Similarly with economists’ models: they can make a lot of sense for prices in a fluid market, they can work OK to model negotiation, they seem like a joke when they start trying to model addiction, suicide, etc.’
I just saw this post. Two distinctions should be borne in mind:
1. Race vs ethnicity. Ethnicity refers generally to human populations distinct in some fashion — culturally, genetically, etc. It really exists. Race is a framework that divides homo sapiens into a very small number of hard categories, typically five or less, based on supposed persistent differences. Modern “scientific” racism defines those differences as genetic, but they have been asserted in more nebulous ways as well. Race is a fictitious category but (of course) no less socially consequential for it.
2. Belief vs validity. Ideology is a theory of the proclivity to believe various propositions as a consequence of one’s social position. (There’s a lot packed into this that I’m not going to try to unpack here.) The key word is “believe”. Ideological belief structures are only indirectly related to “truth value”, via likely flaws in reasoning through arguments, collecting and analyzing data, etc. It’s quite possible for someone to believe X for ideological reasons and for X to be true, although ideologues on average will do a poorer job of sifting evidence about X than those less in thrall. Racism is an ideology; without some argument about proclivity to believe it is difficult to understand how ideas with so little factual basis (see #1) have so much social force. But it’s possible for racists to be right, even insightfully right, about specific propositions or techniques (the eugenic/racist founders of statistics) even though their belief structures were execrable.
Peter:
Point 1 is important and I think it gets confused by a lot of people, including Nicholas Wade and the authors of the Quillette article that someone quoted earlier in this thread. You’ll see people jumping back and forth between different categories of ethnicity to make their points.
1) In public health/medicine sometimes folks only evaluate racial differences, and then sometimes just ethnic differences and sometimes we make a combined variable “racial/ethnic” and use the term racial/ethnic disparities and categorize as White, Black, Hispanic, Asian, Other (if hispanic_eth=1 then race_eth=hispanic regardless if youre Black or White else race_eth=race). There are obviously some issues with this, but a lot of the time it is determined on sample size. If you have 200 patients then you may not be able to make the true number of separate categories to capture variation within categories since you will waste df.
2)Statement: “black infants experience higher mortality rates compared to white infants”
Person 1: I agree with that statement. Blacks are inferior to whites and should die quicker than whites.(I do not endorse this view)
Person 2: I agree with that statement. If systemic unequal race relations (such as unconscious bias/discrimination/racism) were eliminated then the mortality disparity between black and white infants would decrease and perhaps be eliminated entirely.
Whats different? Mechanism. Mediation. Explanation–> Directed acyclic graphs
> Race is a fictitious category but (of course) no less socially consequential for it.
Andrew: Did you really mean to endorse this nonsense? Give Nick Patterson a cheek swab from a random person in the US and he can identify their “race” with 95% (99%?) accuracy. How can something backed up with such hard science be “fictitious?”
If by “race” you mean “set of ethnicities”, then of course by identifying an ethnicity you are identifying the corresponding “race”. This doesn’t make race a meaningful category, however. Moreover, the mapping of ethnicities to races differs from society to society and, as far as I am aware, is biologically arbitrary. I hope this amplification makes my position clearer. If you know of any evidence that undermines it, I would be grateful to hear about it.
Peter Dorman said: “the mapping of ethnicities to races differs from society to society and, as far as I am aware, is biologically arbitrary … If you know of any evidence that undermines it, I would be grateful to hear about it.”
Andrew said: “Race correlates with genotypes and phenotypes”
Lewontin said: 15% of genetic variance is between-race.
Peter: perhaps you should talk to Andrew and Lewontin about the evidence you would be grateful to hear about.
D:
Race correlates with genotypes and phenotypes. But race is also socially determined, as it is a social convention which population differences are labeled as “racial,” which are labeled as “ethnic,” and which are considered not to count at all. This is a point I made in my review of Wade’s book, that the ethnic categories that are given the label “race” vary over place and time.
“race is also socially determined as it is a social convention … the ethnic categories that are given the label “race” vary over place and time”
I don’t understand what conclusions you want us to draw from this statement. Color categories and the naming of dog breeds are also socially determined and social conventions.
> Race correlates with genotypes and phenotypes.
Agreed. Which means that it is not “fictitious.” Andrew, you seemed, above, to be endorsing Peter Dorman’s point 1, which included the claim that “Race is a fictitious category.” As long as we all agree that race is not “fictitious,” then I have no complaint.
Peter Dorman doubles down with “This doesn’t make race a meaningful category.” Again, that is just nonsense. If Nick Patterson can determine your race (with 99% accuracy) from your spit, than race is a meaningful category.
Nick Patterson could, with spit in a tube, also determine (many) Gelman family members vs. Hagen family members vs. Kane family members with > 99% accuracy, so are our different families meaningful categories? Well, sure: Gelman family members might share stories, get together at holidays, speak the same language, and so forth. However, members of my academic department, who belong to different families (and different continental populations), also share stories, get together at holidays, and speak the same language.
“Race” and genes both have very strong essentialist connotations. It is therefore critical to distinguish what genes can tell us about ancestry vs. what genes can tell us about essential traits (biology).
Reich and other geneticists are getting extremely good at determining ancestry from genetic information. The reason is, this problem is (relatively) simple to solve using the increasingly high resolution sequencing data from large samples of many different populations.
Reich and other geneticists still suck, though, at understanding the relationship between genes and essential traits (biology). Example: we’ve known forever that skin color (a key “racial” trait) is heritable, we can measure skin color pretty easily and accurately, we have a pretty good evolutionary story to explain why it varies (variation in exposure to UV radiation), and we know the biological structures involved (e.g., melanocytes). In other words, if there were one trait whose genetic basis should be nailed down by now, it’s skin color. Yet our understanding of the relationship between genetic variation and skin color variation is still very, very fuzzy. We can explain perhaps ~30-35% of the variation. See, e.g.,
An Unexpectedly Complex Architecture for Skin Pigmentation in Africans
https://www.sciencedirect.com/science/article/pii/S0092867417313247
Good points, well put.
(And it makes me wonder how well spit in a tube could be used to identify “Smith” family members with any accuracy — not very well, I think, unless there was a lot of spit from their mothers’ relatives — but, come to think of it, I’ve got Smith ancestry from another line of Smith’s on my mother’s side) ;~)
“Nick Patterson could, with spit in a tube, also determine (many) Gelman family members vs. Hagen family members vs. Kane family members with > 99% accuracy, so are our different families meaningful categories?”
Of course they are. What’s more meaningful and non-arbitrary than biological family trees?
A racial group is merely a special form of an extended family.
You might find my blog post on this issue informative, where I try to disentangle ancestry from biology. Here is the conclusion:
“Each of our many ancestors indisputably lived at certain times and in certain places, and it’s pretty amazing that we can now glean those hard facts from spit in a tube. Living in a certain time and place, however, in and of itself, says nothing about the biology of those ancestors.”
https://grasshoppermouse.github.io/2019/07/27/about-90-of-the-genome-is-junk-which-is-very-informative-about-ancestry-but-says-little-about-biology/
> Living in a certain time and place, however, in and of itself, says nothing about the biology of those ancestors.
You can’t possibly believe this nonsense! Living in, say, 1000 AD Hawaii or 500 BC Norway tells us “nothing” about someone’s biology?! Of course it does!
“in and of itself” is the key phrase here. In and of itself, ancestry says nothing about biology. But coupled with, e.g., environmental factors it can say a lot. Skin color varies in a systematic way with latitude, for instance.
By the way, if you read the linked blog post, you would see that we can determine ancestry using DNA sequences that have little or no influence on the phenotype (biology). In fact, such sequences are ideal for determining ancestry.
Most discussions I see on the topic of race and genetics conflate ancestry with biology. They’re both important, but they’re different.
> ancestry says nothing about biology
Uhh. What can this possibly mean? If you tell me the “ancestry” — meaning the parents or the grandparents or the great-grand parents or whatever — of any organism, then I will know a great deal about the “biology” of that organism.
.. but only if you already know a lot about the biology of the ancestors.
>If you tell me the “ancestry” — meaning the parents or the grandparents or the great-grand parents or whatever — of any organism, then I will know a great deal about the “biology” of that organism.
Only if you know the biology of the ancestors. And if you know that, ancestry doesn’t provide any extra information about the biology.
So yes, we’re getting pretty good at nailing down that someone’s ancestry is Native American vs. European vs. Asian vs. African (or some mix of those). Are there biological differences between those populations? Sure. Do we have anything more than an extremely superficial understanding of what those differences are? No. Do the biological similarities swamp the differences? Absolutely:
https://grasshoppermouse.github.io/2018/06/25/the-universal-genetic-program-and-the-custom-built-phenotype-implications-for-race-and-sex/
The mistake I see a lot of people making when discussing genes and ancestry is that they seem to assume that the ancestry of different parts of the genome is determined by identifying biological affinities and differences. Instead, it is precisely the DNA sequences that have *no* phenotypic effects — those that are pure genetic noise -= that are most informative about ancestry.
> Do we have anything more than an extremely superficial understanding of what those differences are? No.
Depends a lot on what you mean by “understanding.” The smart folks at Genomic Prediction can predict an embryo’s future height to within an inch or two. Does that mean they know the mechanisms by which genes cause height? No! But that sure seems like “understanding” to me.
I thought we were talking about population differences in biology.
It is not obvious (to me) exactly what you are talking about. But you keep saying stuff like:
> we have a very poor understanding of the relationship between genes and phenotypes.
For certain (nonsense?) definitions of “poor understanding,” you might be right. But that specific ignorance is rapidly becoming irrelevant, as the rise of companies like Genomic Prediction demonstrates. If I can predict someone’s height within an inch or two, then my “understanding” is most excellent, at least for a wide variety of tasks which are going to generate hundreds of millions of revenue in the not too distant future.
I’m guessing that you are referencing this study:
https://www.forensicmag.com/news/2018/10/machine-learning-dna-predicts-height-within-inch
https://www.genetics.org/content/210/2/477
So, for example: For people who are 69.1 inches tall, the algorithm predicted somewhere between 67.5 and 70.7 inches just over 50% of the time.
The 25-75th percentile range is 67.2-71.2 inches for US males and 61.7-65.5 for females (both about +/- 2 inches): https://www.cdc.gov/nchs/data/series/sr_03/sr03_039.pdf
So by guessing the median height for a given sex, you would be within 1.6 inches about 50*(1.6/2) = 40% of the time. So their algorithm seems* to be a slight (about half an inch) improvement over just guessing the average height.
Where is the 80% of “hundreds of millions in revenue” for my algorithm of guessing the average?
* They do show concern about checking the performance on a hold-out set (the ARIC data), but after reading over their description a couple times it isn’t really clear to me whether this was only done once at the end or not. So they still may be overstating the performance.
> by guessing the median height for a given sex
If you don’t understand the huge effect that embryo selection is about to have on the world, I have good news! The next decade will provide you with an excellent education.
Reminds me of the gene editing pump and dump. What companies specifically do you see benefiting from this?
> Only if you know the biology of the ancestors. And if you know that, ancestry doesn’t provide any extra information about the biology.
This is nonsense on stilts. The definition of ancestry is “one’s family or ethnic descent.” If I know your ancestry, I know your biology.
Me: “I can predict the future location of a golf ball if I know its current movements.”
You: “Only if you know is mass and velocity.”
Me: “That is what it means to know its current movements!”
You: “But if you know its current movements, then mass and velocity tell you nothing.”
Me: “Arggg . . . “
Me: “ancestry doesn’t provide any extra information about the biology.”
I agree that was poorly phrased.
You: “If I know your ancestry, I know your biology.”
Nope, not without knowing the biology of the ancestors. And we have a very poor understanding of the relationship between genes and phenotypes.
My key point is that ancestry is best determined using DNA sequences that have no effect on the phenotype. Ancestry is one thing, biology is another. Geneticists have made rapid progress in determining ancestry from genetic noise. Understanding the biological implications of DNA sequences that do have phenotypic effects, and the implications of variations in such sequences, is going to be a much tougher slog.
The above is from me. Forgot to enter my details.