Interesting podcast conversation between Sam Harris and Charles Murray (of "Bell Curve" fame)

The BBQ Pit
1541

By request, I’m revisiting the Wiki article. I doubt whether I have much to say that Kimstu hasn’t already covered though. Also, I’m just a jerk looking at wiki page: others on this board have a stronger science background in biology and genetics. Heritability of IQ - Wikipedia

Take the correlation coefficients of identical twins reared apart: that would be an upper bound on the extent that genetics affect intelligence. It’s an upper bound because it’s unlikely that social environments would have been assigned at random: adoption agencies like to assign kids to the best homes they can. Furthermore (and less mentioned in popular discussions) the prenatal environment of the twins are highly correlated, to put it mildly. So yeah, upper bound for the purely genetic effect on intelligence.

Twins reared apart have a Pearson correlation coefficient of .76. That’s not too far from .80, Harris’ reported upper bound.

But that figure is misleading as a guide to the effect of race on IQ. Because genetics aren’t the same as hereditability. To see this, the correlation coefficient between fraternal twins, reared apart. They still share a lot of genetic information - a lot more than 2 people of the same race. And once again, we’re working with an upper bound - post-natal environment will be correlated and prenatal environments will be highly correlated. So what’s this figure?

It’s .35. That’s pretty far from a .50-.80 range.

Now look at biological siblings, reared apart. They will also share a lot more genetic material than 2 people of the same race. The correlation coefficient for them is … .24. And again, it’s an upper bound.

For cousins it’s .15.

Those are pretty low figures - frankly they are lower than I would expect.

Let’s review what high and low correlation coefficients are: What Is R Value Correlation? - dummies
[indent][indent]
0. No linear relationship

+0.30. A weak uphill (positive) linear relationship

+0.50. A moderate uphill (positive) relationship

+0.70. A strong uphill (positive) linear relationship

Exactly +1. A perfect uphill (positive) linear relationship
[/indent][/indent]

…which is to say that .24 and .15 are, to use the technical term, weak. Seriously, that’s the term.

After all is said and done, would it shock me that there was some residual relationship between race (whatever the hell that is) and intelligence (whatever the hell that is)? Brave truth teller MfM says, “No, because why not?” There are lots of subtle and hard to tease out effects in the world. But I’d expect the effect to be small and frankly insignificant in any sort of policy or personal sense. I would also guess that hunter-gatherer groupings would score better than descendants of farmers, because there’s a bigger niche for raw muscle specialists in the latter group. Then again, descendants of urbanites would have awesome disease resistance (and allergies in a modern context).

PS: I’ve dodged an important discussion of R[sup]2[/sup], the correlation coefficient, and other possible metrics for explained variation. Not really the proper forum here.

1542

Not that facts mean anything, but here is a pop-sci article that may make it easier for some people to understand how being poor, or in a minority group that increases stress levels can impact IQ scores.

Of course data won’t convince those that believe some mythical group that they can’t even define is superior.

1543

In terms of pedagogy, there’s some evidence that the black vernacular tends to repress reading scores, partly because of its greater use of homophones. Basically white kids have an edge in sounding out words over black kids who use Ebonics in the home. To see how this plays out in practice, consider this encounter between a linguist and a black four year old:

[INDENT][INDENT][INDENT]Studying African-American Vernacular English wasn’t Julie Washington’s plan. But one day in the fall of 1990, her speech-pathology doctorate fresh in hand, she found herself sitting with a little girl at a school outside Detroit. The two were reading the classic P. D. Eastman picture book Are You My Mother?, which tells the tale of a lost hatchling trying to find its way home. The girl—4 years old, homeless, and a heavy speaker of the dialect known as African-American English, or AAE—listened attentively as Washington read:

[INDENT][INDENT] “Are you my mother?” the baby bird asked the cow.

“How could I be your mother?” said the cow. “I am a cow.”

[/INDENT]
Washington closed the book and asked the girl to recount the story from memory. The girl hesitated, then launched into it. “She goes, ‘Is you my mama? I ain’t none a yo’ mama!,’ ” Washington recalls. “She did the whole thing in dialect.” Washington found the girl’s retelling deft and charming, and she left the classroom smiling.

Only later, sitting in her office at the University of Michigan, did Washington have the flash of insight that would redirect her career. “As a scientist, I stepped back and thought about what that girl had to do,” Washington told me recently, while waiting to address a gathering of linguists at the University of Wisconsin at Madison. “She had to listen to a story in a dialect she doesn’t really use herself, understand the meaning, hold the story in her memory, recode it in her own dialect, and then say it all back to me.” The girl’s “translation” of the book might not sound like much, but translating it? “That’s hard,” Washington said, especially for a young child. The experience convinced her that dialect was playing a significant and unrecognized role in the reading achievement of millions of children—and very likely contributing to the persistence of the black–white gap in test scores. [/INDENT][/INDENT][/INDENT][/INDENT] Julie Washington’s Quest to Get Schools to Respect African-American English - The Atlantic

There ways of addressing this: Oakland tried to take baby steps by treating AAE as a dialect of standard written English, but they were laughed at by ignorant editorialists and the plan was shelved.

1544

Urm, Slacker the original pairing was “Incompetent” and “Incapable”:

1545

Ok, now I’m going to rant a little. In Slacker’s defense, his arguments are substantially less tendentious (i.e. goal directed) than most of the alt-right clowns I’ve encountered on the internet and this message board. Which reflects the fact that Slacker isn’t part of the alt-right.

Still, the big picture is that even following the worst days of Jim Crow (~1948?? IIRC), when educational facilities extended to blacks and whites varied greatly and rewards to black education were diminished, the black white test score gap maxed out at a single standard deviation. Even then, the within group variation was a bigger story than the difference between the means of the 2 groups. And since then the gap has halved.

If you can’t grasp those last two sentences, IM not-so HO, you’re not smart or educated enough to opine on this subject. Yes, they require some parsing. But they are central to the discussion.

Furthermore, IQ wasn’t designed (by Binet) to measure anything intrinsic anyway. It was meant as a way to catch kids that were doing poorly in school, so they could get extra help over the summer. While I’m guessing it is measuring something, I’m not convinced that it’s better than any other standardized test. Which is to say that it’s relationship with eg lifetime earnings or any other direct outcome is a loose one.
Most alt-right clowns don’t like to dive into these complexities. In fact, the concept of standard deviation either eludes them entirely or at the very least fills them with indifference, disinterest. Despite the fact that it’s core to their position. Bloody annoying shitheads don’t have a clue how messy the underlying science is, and don’t care.

1546

This is fascinating stuff, and I absolutely understand why one could use this to make a strong argument against there being a significant racial-genetic component to IQ (although one thing I wish were included in your figures as a kind of control is what the correlation coefficient is between two randomly chosen adopted kids, who are not only “reared apart” but would have similar environments in terms of being chosen by adoption agencies, yet not even be cousins in terms of biological relatedness). We would expect it to be more than zero, given the similar environments, right?

But there’s something that jumps out at me that I just can’t wrap my mind around, and I wonder if you know what to make of it or have any hypotheses. (Let me note in advance that I’m not lawyering for a genetic race-IQ argument here, just purely following the data and trying to understand it.) And that’s the sharp dropoff between the identical twins and the fraternal twins. You note that the home environment of adopted kids is not random, due to placement by adoption agencies. And also that of course prenatal environment is shared by fraternal twins, just as it is by identical twins. So what else can that huge .41 dropoff represent, except for the genetic difference between identical and fraternal twins? Doesn’t that mean getting just the right combination of genes is the dominant factor in IQ, leaving race aside?

And yet apparently even “identical” twins don’t share the exact same genomes:

Which means .76 is not necessarily the upper bound after all–just ask that poor twin who has leukemia. So that huge .41 dropoff not only has to be all about genetics (unless I’m missing something), the actual dropoff–if identical twins were truly identical genetically–might even be higher!

Isn’t this almost a paradox? While the weak correlation you get by the time you are at the sibling, and especially the cousin, level seems to make the genetic hypothesis at the race level dubious, we do in fact have a large disparity in IQ scores between races, even at young ages and even when comparing poor black kids with poor white kids or rich black kids with rich white kids. So given that there’s this huge chunk of correlation gobbled up just by the genetic difference between fraternal and “identical” (one-egg) twins, is there room enough left for “nurture” to explain those racial disparities?

Something doesn’t add up here. Or if it does, please show me how!

Yes, and I don’t agree that those are quite synonymous, which is why I said “not quite”. GIGO responded with a cite that also called “naive” a synonym of “incapable”, which to me demonstrated how loose that source was being in its definition of “synonym”. Meaning that it didn’t really help GIGO’s case—and I feel like I’ve seen him/her undermine his/her own arguments before, in the midst of ranting and raving at me (which you might notice I don’t do in response). That’s all, not a huge deal.

Yeah, I remember that, and I agree that it’s a shame. I think they made a mistake calling it “ebonics” instead of AAVE (I’ve always seen it with the V, but I see you eschew that).

It’s a good point, that if we treat black kids as semi-ESL speakers, that might really change the results of a lot of aptitude tests. It still doesn’t account for smaller cranial size or for the poorer performance of black kids adopted by white families, though.

In any case, my periodic disclaimer: if we stop as a society throwing educators and teacher unions under the bus for lower test performance by black students compared to whites and Asians, I’d be delighted to drop this whole distasteful subject and frown at anyone who seems to relish digging into it. I don’t want to shame anyone unnecessarily, which I think is the biggest difference between me and the alt-right (and I appreciate your acknowledging that I’m not in that category).

1547

To be honest, my main takeaway was that these relationships are too complicated to address with simplistic thinking or simplistic statistics. Never mind tendentious bigotry, which I trust you’ve observed on this message board as well.

Yes, more data is better.

Also, generally speaking a paper that just used correlation coefficients wouldn’t be publishable. You want to know what the correlations are after controlling for this or that. In other words you want to multi-variable approach. Ordinary least squares - Wikipedia (Yup: “Ordinary”)

It isn’t unusual for tables of cross-correlations to have stuff that jumps out. When I see a dataset like that (i.e. when I see basically any dataset) I typically stop screwing around with the correlation coefficients, and work on some sort of multi-variable model. If you have the raw data, this is easier than it may sound - Excel has had an Ordinary Least Squares plugin since at least the early 1990s.

Didn’t know that. That does change my upper bound argument.

I find this to be common when you dig into data. While it sometimes is roughly consistent with my pre-conceptions, there are almost always twists of one sort or another. Then again, I’ve once heard data investigation as a process of learning more and more about less and less.

Holy shit, cranial size? That’s 1800s stuff.

You know what cranial size is correlated with? Strongly? Body size. No established relationship with intelligence at least c. 1981. Cite, this book: The Mismeasure of Man - Wikipedia

(Yes, I see that the wiki article contains some back and forth regarding the effects of unconscious bias on measurement of cranial size. Doesn’t affect my claim. )

1548

According to this Stanford neuroscientist, brain size does matter, even if it’s not everything (why this would even need to be stated, I can’t imagine): Ask a Neuroscientist: Does a bigger brain make you smarter? | Wu Tsai Neurosciences Institute

1549

Putting this here because it’s kind of personal. But I think this, once again, belies your earlier claim that you don’t care about credit. You want credit, or reassurance, that no, you aren’t so bad, because you liked Obama and reparations and stuff like that.

But that’s not how it works. Racism and bigotry in society and culture are incredibly insidious forces that affect all of us in various ways, and until you (and folks like Sam Harris) consider and don’t outright reject the possibility that your beliefs, especially in things like inherent genetic intellectual inferiority and superiority, could be influenced by these insidious and powerful societal forces, you won’t get any credit or reassurance that you’re on the right side of this issue, at least not from me. I’m glad you’re not a hateful person, but making progress requires a lot more than an absence of hatred.

1550

That was a meta study, that has huge caveats called out in the source data. And it is a perfect example of the logical error that you seem to keep making. I am not blaming you as confirmation bias is a challenge for all humans but remember that correlation does not equal causation.

From the same era as your study.

And more recent data has demonstrated it is far more complex.

https://www.nature.com/articles/nn.2135
This is why the scientific method uses techniques like testing the null, as to avoid this type of cherrypicking.

It is also why good scientists are just about as happy when they get a result that disproves their hypothesis.

This is the difference between a scientist like Brian Greene would be extremely excited if you could disprove string theory; where Deepak Chopra will just find another tortured description to justify his beliefs or dismiss what you said.

You are still Begging the question.

1551

As **iiandiii **notices, you are not just reminding all about your cowardice but here you are also reminding us about your insecurities. That was just about the fourth time where you worry about the grammar of a poster rather than the information that experts provided about how wrong you are about the main issue. And worst of all: that was a very, very weak reply about the grammar issue too.

So, good thing that I’m not a coward regarding the main issue because as others noted, you are cherry picking.

Leave it for the racists to miss the elephant in the room.

As for cranial size and IQ differences:

http://bactra.org/sloth/lieberman-on-rushton.pdf

1552

Andy, it doesn’t “belie” anything. I said all along that I don’t like my views to be misrepresented. Lumping me in with people who bemoan the Great Society, hate Obama, snarl at minorities, or pine for the “Lost Cause” is a gross misrepresentation.

GIGO, that’s interesting about the earlier claims that Asians had smaller heads. But it looks pretty obvious to me that those earlier measurements were distorted by racial bias and the modern ones, using MRIs, are accurate. What racial agenda of the 21st century would require falsely putting Asians on the top level?

1553

They’re not being misrepresented, and you’re not being lumped in with anyone, you’re just being criticized for things that you’ve said.

1554

You are still missing the main point here.

https://www.sciencedirect.com/science/article/pii/S0191886909003675
There is no prove that this correlation is causal, and in fact the grey vs white matter ratios and other internal, non-volume related features seem to be more related to IQ tests (which is not a reliable measure of intelligence BTW).

Note that most differences in volume are explained by height and the effect is tiny

Brain structure mediates the association between height and cognitive ability.

You are accepting very weak data based purely based on confirmation bias. And you are over representing the differences that these studies even suggest might be true.

1555

That’s horseshit, but I have corrected the record.

Rat avatar, funny you mention confirmation bias. I have a sneaking suspicion you’d find a study with results of the same magnitude much less “weak” if it said what you wanted it to say. :dubious:

1556

Perhaps, but when you provide cites that call out the weakness of current studies as a “smoking gun” in support of your racial superiority claims the likelihood that I will be exposed to information that will change my mind is nil.

Or did you not read your entire cite at this URL? NeuWrite West -- Ask a Neuroscientist: Does a bigger brain make you smarter?

Also note how you are ducking the cites I offered debunking your claims that weren’t even supported by your cites in the first place but you still won’t address them.

I have tried to address every point you have made, and you have ignored mine…I will stand on my position until you actually can provide evidence for your baseless, racist claims.

1557

Pish and tosh. I acknowledged the strength if Measure’s point. You never concede anything.

1558

I am not Measure, and I have nothing to conceded to until you actually define what “white” is, which you refuse to do.

I can address the problems with your claims for white supremacy, but I cannot concede that something is possible when you are unwilling to define your base assumptions.

If you stated that people with normal lotus-o-deltoid type earlobes were less intelligent than those with semi-boloid earlobes how could I agree if you were unwilling to define what the difference was between lotus-o-deltoid and semi-boloid earlobes?
You are simply begging the question, and refusing to address that point while making more and more outrageous claims.

1559

Upthread I cited a large study, again from Stanford, that stated genomic groupings corresponded to self-reported race to an uncanny degree of accuracy. That wasn’t good enough for you, because nothing is ever good enough for you. You are a true Scotsman, sir!

1560

You ignored my challenges to that cite, which selected a small number of “races”, and ignored areas of overlap, and the fact that of the “races” they studied, 50% weren’t even a “race” in the first place but an ethnic group, which the study itself calls out.

So tell me, Are Jewish people white? or are they Asian? if this is your definition, How about people from Arabic countries or from India? As they had a small sample size for Africans where do the Egyptians fit on your model.

This study did NOTHING to show that race is a reliable genetic marker, only that when you cut down your sample size to extremes you can detect very coarse differences, but not for Hispanic people.

Look at the goal of the study.

And note how they call out that the assumption you are making to support your white supremacist claims is called out as an assumption that requires verification.

Maybe it will help if you read the studies vs linking to a pop-sci news letter?