More Race and PC problems at Harvard

[Stalker]

Well, anyway, I'm sure in a few years DNA sequencing and analysis will be cheap and easy enough that researchers will not have to use skin color as a marker and can examine specific alleles directly. Then we can forget about this stupid debate.

I assume you mean populations will be tailored based on DNA, so IQ testing would be done relevant specific alleles? Testing subpopulations, or overlapping populations, would only be interesting if they could be convinced that they don't already know the distributions of the relevant genes.

So is class

Marxist class in an objective economic relationship, not a social construct.

A. That the (discrete) races differ in cognitive ability (IQ tests, SAT, reaction time, etc.) is a fact that almost nobody disputes. The reason for this fact is an open question.

The problem with statement A is that it is framed using discrete races. This makes it difficult to communicate, because the framework itself pushes some buttons. So let’s do better.

Buttons like intellectual honesty? (I can play that card too.)

B. Groups differ in cognitive ability (IQ tests, SAT, reaction time, etc.) correlating to the degree of recent sub-Saharan ancestry. This is a fact that almost nobody disputes. The reason for this fact is an open question.

The reference I provided was to support the factual (and interesting) claim. [...] Interesting stuff, and (from where I’m standing) an obvious field to research, now that we can genome-sample your degree of African ancestry much better than you can self-identify it.

Is it? Interesting, useful? Why is it more interesting than, say;

C. Groups differ in cognitive ability (IQ tests, SAT, reaction time, etc.) correlating to the degree of recent ancestry from the Asioafrican population. This is a fact that almost nobody disputes. The reason for this fact is an open question.

The studies you linked would equally support that, and if the populations chosen are arbitrary, why not make that choice?

[Emo Targaryen]

more often than not, people (and a tree) will pop up with claims stating that there's a genetically determined IQ gap between whites and blacks and Jews and Asians. So I then feel compelled to comment on the inaccuracy of that framework.

I wouldn't talk of a "gap". It's not like you could separate people into "IQ < 90 -> black; 90 < IQ < 120 -> white; IQ > 120 -> jewish". But, while I don't feel in a qualified position to claim any real knowledge on this topic, I don't categorically rule out the possibility that, if you grouped people inside a society or even the whole world into "races" following however arbitrary criteria, there could be a statistically significant correlation between this classification and a however arbitrary measurement of "intelligence". This doesn't mean the "race" of an individual would be any reliable test to determine its "intelligence"; but if you measure "intelligence" and get statistically significant different averages for different "races", you cannot a priori claim this is impossible and must be due to flawed methodology. If you do so, you are embracing a non-falsifiable theory as defined by Karl Popper and thus are placing yourself outside the scientific community, along with creationists, marxist "historical materialists" and freudians.

@Aemon Stark:

Well, there's no evidence that "race" is a causal predictor of anything. That it's correlated with a number of different outcomes doesn't say very much.

Well, that it's correlated does say that it has to be linked either by a direct or indirect causal relationship in one or another direction or by a common cause. Which one of those applies, can be deduced from statistical analysis. However, a "causal relationship" says nothing about the individual. There's a strong and widely undisputed causal relationship between smoking and lung cancer, but there's still a lot of heavy smokers with no signs of cancer and a lot of non-smokers with cancer. Accordingly, even if a (much weaker) causal relationship between race and intelligence exists, there's no reason whatsoever to expect stupidity from a black person or brilliance from a Jew.

I think the reason why such a simple statement causes that much uproar is the following: a lot of current political action is based on the assumption that any correlation between race and certain outcomes (income, education, crime) has to be caused by racism. Any claim of any genetic explanation (or, for that cause, any social explanation different from racism) would undermine the raison d'être of Affirmative Action and the like.

[Aemon Stark]

I wouldn't talk of a "gap". It's not like you could separate people into "IQ < 90 -> black; 90 < IQ < 120 -> white; IQ > 120 -> jewish". But, while I don't feel in a qualified position to claim any real knowledge on this topic, I don't categorically rule out the possibility that, if you grouped people inside a society or even the whole world into "races" following however arbitrary criteria, there could be a statistically significant correlation between this classification and a however arbitrary measurement of "intelligence". This doesn't mean the "race" of an individual would be any reliable test to determine its "intelligence"; but if you measure "intelligence" and get statistically significant different averages for different "races", you cannot a priori claim this is impossible and must be due to flawed methodology. If you do so, you are embracing a non-falsifiable theory as defined by Karl Popper and thus are placing yourself outside the scientific community, along with creationists, marxist "historical materialists" and freudians.

The problem with arbitrary criteria is that they're arbitrary. Unless racial groupings give rise to discrete, independent populations, you cannot use simple comparisons of means to make any statistical inferences. But racial groups are neither discrete nor are they non-overlapping; to the extent the concept makes sense as anything but a form of self-identification, such groups are at best clusterings of a handful of physical traits that are considered important to that identity. Of course, these physical traits are themselves highly variable. Sure, you can compare means of these different groups and comment on how many standard deviations they lie apart from one another, but it's not meaningful to claim statistically significant differences between groups that are not truly independent.

@Aemon Stark:

Well, that it's correlated does say that it has to be linked either by a direct or indirect causal relationship in one or another direction or by a common cause. Which one of those applies, can be deduced from statistical analysis. However, a "causal relationship" says nothing about the individual. There's a strong and widely undisputed causal relationship between smoking and lung cancer, but there's still a lot of heavy smokers with no signs of cancer and a lot of non-smokers with cancer. Accordingly, even if a (much weaker) causal relationship between race and intelligence exists, there's no reason whatsoever to expect stupidity from a black person or brilliance from a Jew.

I think the reason why such a simple statement causes that much uproar is the following: a lot of current political action is based on the assumption that any correlation between race and certain outcomes (income, education, crime) has to be caused by racism. Any claim of any genetic explanation (or, for that cause, any social explanation different from racism) would undermine the raison d'être of Affirmative Action and the like.

No, correlation doesn't imply anything about causation. None of these studies have any degree of randomization, whether at the level of study design or subject recruitment. They are purely observational and I doubt very much that they conform to any epidemiological standards. All that aside, the lung cancer example is a good one. The link between lung ca. and smoking depends on a very well elucidated pathophysiology and a dramatic risk association. The nature of the data is further entirely different - we can actually define discrete groups, e.g. people who have never smoked regularly, people who have smoked regularly but not in the past 5 years, people who smoke regularly now, etc. We can even simply define dose in terms of pack years. The outcome variable is simple and categorical - lung ca. or not.

Incidentally, though, smokers comprise the vast majority of lung cancer patients, around 80%, and the remainder of cases can be attributed to certain environmental or occupational exposures with only a few occurring sporadically. But smoking is still just the most important risk factor for lung cancer; the relationship is not deterministic despite being far more robust than any of this intelligence research.

[IheartTesla]

I don't know about that. I've never heard anyone talk about any other factor besides skin color. Or maybe hair?

Maybe I don't follow you here very well.

I thought all I was saying is that skin color =/= race. But my puny physicist brain cant handle all these correlations floating around, and I'm not much of a biologist, so maybe I'm just talking out of my ass.

My final thoughts are that IQ differences are hardly static over time, and there is some merit in the notion that many generations of slavery and segragation are at least partly to blame for the current IQ gap between blacks and whites. And Pinker agrees with me, I can quote relevant sections from The Blank Slate. So to think that these numbers will remain the same in say a hundred years or so requires a level of hubris I am not comfortable with.

Everything is transient....

[Emo Targaryen]

Unless racial groupings give rise to discrete, independent populations, you cannot use simple comparisons of means to make any statistical inferences. But racial groups are neither discrete nor are they non-overlapping; to the extent the concept makes sense as anything but a form of self-identification, such groups are at best clusterings of a handful of physical traits that are considered important to that identity.

Well, in this case any quantitative sociology is moot, because social strata or "classes" are just as overlapping and ill-defined as races. The point is, if there are any criteria according to which you can group individuals into classes without any a priori knowledge of their intelligence, and then get different IQ means for these classes, these criteria and intelligence are correlated. Refined research could then make the bins smaller (classify by degree of recent sub-saharan ancestry, by degree of self-identification with the black community, by degree to which people are percieved as black by others) and see along which axes most correlation occurs.

it's not meaningful to claim statistically significant differences between groups that are not truly independent.

What do you mean by "truly independent"? Mutually exclusive groups are by definition not independent in a statistical sense, rather the opposite.

No, correlation doesn't imply anything about causation. None of these studies have any degree of randomization, whether at the level of study design or subject recruitment.

I haven't looked closer at any of these studies - do you mean to say they all intentionally are comparing black prison inmates to white professors? Be that as it may, while correlation doesn't imply direct causation, it does imply something about causation. If two things are absolutely independent, they aren't correlated.

The link between lung ca. and smoking depends on a very well elucidated pathophysiology

But that's not the point in statistical research. Even if we were completely clueless about the way of causation, from the statistical link we would have to assume some form of causation. Either a direct causation ("smoking causes cancer" or "having cancer makes people smoke"), an indirect causation ("smoking costs a lot, so smokers thend to shop for cheaper items which contain more cancerogenic substances") or a common cause ("gene AZX46 has two effects: it increases cancer probability and it increases the addiction potential of nicotine") or a mixture of all these.

The nature of the data is further entirely different - we can actually define discrete groups, e.g. people who have never smoked regularly, people who have smoked regularly but not in the past 5 years, people who smoke regularly now, etc. We can even simply define dose in terms of pack years. The outcome variable is simple and categorical - lung ca. or not.

We can also define discrete groups of people whose skin is lighter than a brown paper bag and people whose skin is darker. We can even simply define skin color in terms of coordinates inside an RGB or HSV color space.

the relationship is not deterministic despite being far more robust than any of this intelligence research.

Of course it isn't. As the link between race and intelligence, if it exists at all, surely isn't. I don't see anyone claiming that.

[Happy Ent]

Is it? Interesting, useful? Why is it more interesting than, say;

C. Groups differ in cognitive ability (IQ tests, SAT, reaction time, etc.) correlating to the degree of recent ancestry from the Asioafrican population. This is a fact that almost nobody disputes. The reason for this fact is an open question.

The studies you linked would equally support that, and if the populations chosen are arbitrary, why not make that choice?

Well, we'd need to tinker with population sizes, but for the sake of arguments let's say this works. Blasians. They'd have higher IQ than Whites, and some of the variance would be well explained by recent ancestry. Just as you say.

The reasons we don't do this are:

(1) there is no currently popular social category called Blasians. There sure used to be ("bloody furriners"), but currently there isn't.

(2) Blasians doesn't make a lot of sense ancestry-wise, in fact, even less than "Asians" or "Blacks". It's a step in the wrong direction and would probably produce results with very little statistical significance. It's even more arbitrary. (Your argument is another good example of being caught in dichotomic thinking: "If something displays some degree of arbitrariness, then all things are equally arbitrary.")

(3) Most importantly, if you did this (for example, if the US had only two US Census categories: WASPS and Everybody Else) then affirmative action would die. So it will never happen. The US Census categories are what they are exactly because there's a strong, benevolent political movement behind them. As soon as you introduce Blasians, all the indicators for racism will evaporate, because Blasians do at least as well as Whites.

[Emo Targaryen]

Well, we'd need to tinker with population sizes, but for the sake of arguments let's say this works. Blasians. They'd have higher IQ than Whites, and some of the variance would be well explained by recent ancestry. Just as you say.

It would however not be a very sensible category as we could split them into two groups ("African" and "Asian") with different intelligence results. I'd even say it would be obvious that there must be the possibility of such a division, because you would see two distinct peaks if looking at the "blasian" intelligence distribution.

Likewise, we could e.g. split Blacks into two groups, based on whether their ancestors came to America in chains or as free men (like Obama's father) and see if those groups behave differently intelligence-wise.

[Stalker]

We can also define discrete groups of people whose skin is lighter than a brown paper bag and people whose skin is darker. We can even simply define skin color in terms of coordinates inside an RGB or HSV color space.

Yes, but this is not what has been done. Inclusion criteria have varied, based on the assumption it is enough that they identify race. When that is false, would you agree that this means they have been testing different (but correlated) populations and hen treating it like they've tested the same population?

There is also the question why testing one arbitrary population would be more interesting and useful that another arbitrary population.

[Emo Targaryen]

When that is false, would you agree that this means they have been testing different (but correlated) populations and then treating it like they've tested the same population?

If anyone is treating it that way, it's us or researchers compiling a meta-research. But yes, this means data from different research projects is not necessarily comparable, as they tested slightly different populations (so it cannot be used e.g. to plot the evolution of the results along time, as research from the 30s will probably work with different populations than research from the 90s). It does however not mean that this data is utterly useless and tells us absolutely nothing.

There is also the question why testing one arbitrary population would be more interesting and useful that another arbitrary population.

If arbitrary criteria A show more correlation to a certain set of factors than arbitrary criteria B, this means it's more useful to differentiate along criteria A if talking about said factors. E.g. there's a correlation between sex and sprint performance and that's why there's separate contests for male and female sprinters at the olympic games. At least that's a statistician's viewpoint. A sociologist might even say that criteria A are better suited to differentiate society into classes, races, strata or whatever.

[Aemon Stark]

Well, in this case any quantitative sociology is moot, because social strata or "classes" are just as overlapping and ill-defined as races. The point is, if there are any criteria according to which you can group individuals into classes without any a priori knowledge of their intelligence, and then get different IQ means for these classes, these criteria and intelligence are correlated. Refined research could then make the bins smaller (classify by degree of recent sub-saharan ancestry, by degree of self-identification with the black community, by degree to which people are percieved as black by others) and see along which axes most correlation occurs.

I'm not clear on how grouping people by income quintiles (or something similar) gives rise to overlapping groups; they are certainly discrete with respect to socioeconomic status. Similarly, racial groups are discrete insofar as people self-identify as belonging to one race or another (though where does that leave Obama?), so as social groups they are non-overlapping. This is not the case with respect to "biological" criteria, however, so inferences made that assume these social groups exactly coincide with meaningful, discrete biological groups are not valid.

Concerning nonsense like "recent sub-Saharan ancestry", HE has done nothing but made a lot of unsubstantiated assertions in that respect. It is further entirely reasonable to doubt that African Americans are in particular representative of the current inhabitants of sub-Saharan Africa.

What do you mean by "truly independent"? Mutually exclusive groups are by definition not independent in a statistical sense, rather the opposite.

Not really statistical independence, though I should be using mutually exclusive in this situation more technically. They are not opposites, though certainly not synonymous either.

I haven't looked closer at any of these studies - do you mean to say they all intentionally are comparing black prison inmates to white professors? Be that as it may, while correlation doesn't imply direct causation, it does imply something about causation. If two things are absolutely independent, they aren't correlated.

No, I'm saying that these studies are non-randomized, not randomly sampled, and not matched for sociodemographic variables. They are purely observational, which doesn't tell us nothing, but it doesn't provide much more than descriptive interpretations. It is entirely normal to obtain non-zero correlations between two variables; we are only looking for sufficient evidence to violate the assumption of independence. Correlations can be and frequently are spurious, non-significant, or artifacts. In this case, we have only one continuous variable and so the type of correlations you seem to be discussing is not available.

But that's not the point in statistical research. Even if we were completely clueless about the way of causation, from the statistical link we would have to assume some form of causation. Either a direct causation ("smoking causes cancer" or "having cancer makes people smoke"), an indirect causation ("smoking costs a lot, so smokers thend to shop for cheaper items which contain more cancerogenic substances") or a common cause ("gene AZX46 has two effects: it increases cancer probability and it increases the addiction potential of nicotine") or a mixture of all these.

I don't think we're really disagreeing here, but there's nothing especially dramatic about the apparent group differences in intelligence. I was commenting on the strength of the association, and the flimsy evidence that it has anything to do with "innate" or "genetic" differences. That these claims are made by certain psychologists with connections to eugenics-based organizations should give us pause. With smoking, at least, we have lots of well documented and research pathophysiologic changes to explain *why* the association exists. Correspondingly, we know that environment is significantly determining of a host socio-economic, health, and occupational outcomes, and that the interaction of genes and the environment in which they're expressed plays a significant role in phenotype. The dichotomy wherein it's either "heredity" or "culture" is antiquated and, I will grant, ideological, but that extends in either direction.

We can also define discrete groups of people whose skin is lighter than a brown paper bag and people whose skin is darker. We can even simply define skin color in terms of coordinates inside an RGB or HSV color space.

So we replace one arbitrary definition with another? There is actually some controversy in epidemiology about the choice of "cut points" between low/high exposure to a disease-causing entity, since the choice significantly determines the apparent strength of the results.

[Tempra]

It appears that Raidne and FLoW were correct. This conversation began with a discussion of affirmative action.

The conversation turned to affirmative action. Steph and Yelena oppose it; Jen is in favor. The three had the kind of argument that all of us have probably had, in college or law school — spirited, engaging, passionate. There was some adoption of perhaps extreme or exaggerated views, to get a rise out of the other participants, in the spirit of law school intellectual bravado.

During the dinnertime debate, Steph did not argue in favor of a genetic basis for racial disparities in intelligence. After the dinner, however, she sent an email — just to Yelena and Jen, not a wider group — clarifying her views. In that email, Steph wrote, “I absolutely do not rule out the possibility that African Americans are, on average, genetically predisposed to be less intelligent.” (For the few of you who haven’t already seen it, you can read the complete email here (third blockquote).)

Note the wording of Steph’s email: “I absolutely do not rule out the possibility….” This suggests that, at dinner, Steph actually argued against racial disparities in intelligence. Upon further consideration, she decided to go agnostic on that question, sending out the clarifying email. For a layperson without expertise in the relevant scientific disciplines, agnosticism on this subject seems reasonable — and does not make someone a “racist” (not that Kash, in our initial post on this subject, called Steph a racist, even though Kash believes the email contains racist subtext).

http://abovethelaw.com/2010/05/the-harvard-email-controversy-how-it-all-began/#more-15243

[TerraPrime]

For a layperson without expertise in the relevant scientific disciplines, agnosticism on this subject seems reasonable — and does not make someone a “racist”

It is not "reasonable," in my worldview.

[Ran]

Hrm. I find it difficult to see how agnosticism on the question is an unreasonable position to take, to be honest. While I strongly believe cultural and environmental factors explain pretty much all differences across average populations (including factors we just don't really have a grasp of), I've never seen a smoking gun that 100% proved this was absolutely the case, and so it's not far-fetched to suspend judgment.

[Lany Freelove Cassandra]

I have discussed many "interesting" things in e-mails with friends with whom I enjoyed debating. There were times when I even played devil's advocate in those discussions. My views also used to swing a bit to the right, but were certainly never racist. At least I never thought they were. I am just wondering how I would feel if it was my views, or worse yet, statements I made in private as part of a debate that ended up public and out of context and I was ostracized in my community. (actually, it's more my lefty views that would cause that where I live, but the point is still the same)

First, I would feel very betrayed that a private conversation was made public. Second, as a quirk of my personality, I would feel I needed to apologize and explain myself...both things that would likely make the situation worse.

It doesn't matter if her OPINIONS are right or wrong (or even left) but that her privacy was violated, her trust betrayed and now people what her whole career to be destroyed because of a PRIVATE CONVERSATION. This is what is wrong.

Why the hell should anyone care what this one woman said in a discussion about anything really? Maybe if one day she was going to be a superior court judge it might matter (but jeez, something she said 20 years ago while still in school is still haunting her?).

[Happy Ent]

No, I'm saying that these studies are non-randomized, not randomly sampled, and not matched for sociodemographic variables.

I’m not quite sure what kind of study you are soliciting here. Certainly, sociodemographic values are controlled for in many studies.

From he Rushton–Jensen survey:

Some culture-only theorists propose that SES should not be assessed in terms of crude material measures but must be seen as a complex of attitudes, aspirations, self-images, and societal stereotypes (Loury, 2002; Ogbu, 2002; Sowell, 1994). Some of these types of cultural factors have been tested as well. Matching Black and White children for the geographical areas of their homes, the schools they attend, and other finer grade socioeconomic indicators again reduces the mean group IQ difference but does not eliminate it. Black children from the best areas and schools (those producing the highest average scores) still average slightly lower than do White children with the lowest socioeconomic indicators (Herrnstein & Murray, 1994, pp. 286–289; Jensen, 1998b, pp. 357–360). This is an anomaly for the culture-only theory but is explained by genetic theory through regression to the mean (see Section 10).

Regarding SES, for people who like to look at graphs, the bottom of this page at Griffe du Lion shows SAT achievement broken down by race and variables such as parental education and income. I find these charts jaw-droppingly terrible, they make my eyes burn and my heart break. Controlling for these factors does not seem to help the environment-only side. (That being said, I’m not in favour of having this debate on the basis of a single study or data point.)

(Aemon, you seem to think “environment-only” is a strawman for a position that nobody holds. I’d love this to be true—it is an artefact of these kinds of debates the the side representing the orthodoxy seldom needs to make positive claims like “The observed gap between White and Asian IQ is entirely due to XYZ,” and can fall back on Crimestop instead. But I see that several lucid board members have actually come out with statements that are concordant with the environment-only hypothesis, so forgive me for keeping the position alive.)

[Emo Targaryen]

I'm not clear on how grouping people by income quintiles (or something similar) gives rise to overlapping groups; they are certainly discrete with respect to socioeconomic status.

Yes, but income quintiles are very arbitrary. Compare your own words:

There is actually some controversy in epidemiology about the choice of "cut points" between low/high exposure to a disease-causing entity, since the choice significantly determines the apparent strength of the results.

They're just lines drawn at mathematically convenient places. If you want to define something like a "working class", you can't just go by quintiles and you have an unclearly defined border to other groups like "proletariat" or "middle class".

Similarly, racial groups are discrete insofar as people self-identify as belonging to one race or another (though where does that leave Obama?), so as social groups they are non-overlapping. This is not the case with respect to "biological" criteria, however, so inferences made that assume these social groups exactly coincide with meaningful, discrete biological groups are not valid.

But why would they have to exactly coincide? Yes, there are some people who would be grouped into one race by one definition and into another race by another definition. As long as working with both definitions gives roughly the same result, we can infer that these individuals have no significant effect and thus it doesn't matter where we classify them (remember, we're only classifying them for statistical purposes here. It doesn't make any difference to them where we classify them). If results differ, we can analyze which definition gives the clearest results.

It is further entirely reasonable to doubt that African Americans are in particular representative of the current inhabitants of sub-Saharan Africa.

As it is reasonable to doubt that they are statistically significantly different from them. We should wait for someone to do genomic analysis and standardized intelligence tests all over sub-saharan Africa to settle this question.

In this case, we have only one continuous variable and so the type of correlations you seem to be discussing is not available.

What do you mean "not available". It's all the more available as there is as yet no statistical evidence to rule it out. The answer to bad research is better research - and as long as there's none of this, there's no reason to categorically rule out one or the other causal explanation or even to denounce people as "racist" for not ruling out (not even embracing) one of these possibilities.

That these claims are made by certain psychologists with connections to eugenics-based organizations should give us pause.

No, it shouldn't. Not any more than any new results, no matter who obtained them, should give us pause and make us check for the methodology they used. It's a popular tactic nowadays to denounce unwanted results by pointing out economical or political affiliations of the researchers - and thereby inferring that those who obtained the opposite results are pure idealists, not driven by any agenda but only by the quest for the holy grail of scientific truth. Science doesn't work that way. Science has no political litmus tests, science has scientific standards and as long as research keeps to them, its results should be considered legit no matter what corrupt nutjob the actual researcher might be.

Correspondingly, we know that environment is significantly determining of a host socio-economic, health, and occupational outcomes, and that the interaction of genes and the environment in which they're expressed plays a significant role in phenotype.

Sure. But this means we can start to research the degree to which nature and nurture respectively determine the outcome and what genes or environmental factors are most strongly correlated with the outcome.

[Aemon Stark]

As it is reasonable to doubt that they are statistically significantly different from them. We should wait for someone to do genomic analysis and standardized intelligence tests all over sub-saharan Africa to settle this question.

And we'd be doing this why, exactly? To disprove the claim that Africans are less intelligent on average than other groups? The very question is implicitly racist and I see zero indication that the motivations behind this research derives from anything else. Similar arguments are made now for why Haiti, to take one example, is a "failed state".

What do you mean "not available". It's all the more available as there is as yet no statistical evidence to rule it out. The answer to bad research is better research - and as long as there's none of this, there's no reason to categorically rule out one or the other causal explanation or even to denounce people as "racist" for not ruling out (not even embracing) one of these possibilities.

"Not available" in the sense of not able to be calculated.

No, it shouldn't. Not any more than any new results, no matter who obtained them, should give us pause and make us check for the methodology they used. It's a popular tactic nowadays to denounce unwanted results by pointing out economical or political affiliations of the researchers - and thereby inferring that those who obtained the opposite results are pure idealists, not driven by any agenda but only by the quest for the holy grail of scientific truth. Science doesn't work that way. Science has no political litmus tests, science has scientific standards and as long as research keeps to them, its results should be considered legit no matter what corrupt nutjob the actual researcher might be.

I don't know what this "science" is that you're talking about, but research ethics requires disclosure of funding sources and declaration of conflicts of interest. Research funding from, say, a pharmaceutical company does not inherently bias the research outcomes by any stretch, but the regulations for such research are considerably more stringent and monitored than for the average peer-reviewed journal. If the Lancet can get duped into publishing things like Wakefield's autism and MMR vaccine "study", then some degree of skepticism is inherently justified. Rushton himself currently heads the aforementioned Pioneer Fund. Whatever their stated mandate, they seem to be rather heavily involved in funding "research" that conforms to the racist beliefs of their founders.

Sure. But this means we can start to research the degree to which nature and nurture respectively determine the outcome and what genes or environmental factors are most strongly correlated with the outcome.

No, it's not just a matter of additive interactions between genes and environment - the two effects are themselves correlated. It's all well and good to attempt to discern the independent linear effects of each separately, but that only imposes a statistical model which is a gross oversimplification.

[Datepalm]

Regarding SES, for people who like to look at graphs, the bottom of this page at Griffe du Lion shows SAT achievement broken down by race and variables such as parental education and income. I find these charts jaw-droppingly terrible, they make my eyes burn and my heart break. Controlling for these factors does not seem to help the environment-only side.

How? They show that children of black holders of graduate degrees do badly. Are blacks really bad at passing on smart-genes or are black graduates just pretty dumb? (those degrees were probably earned with affirmative action anyway.) The Asian lousiness at verbal components seems a bit suspect too.

[Former Lord of Winterfell]

How? They show that children of black holders of graduate degrees do badly. Are blacks really bad at passing on smart-genes or are black graduates just pretty dumb? (those degrees were probably earned with affirmative action anyway.) The Asian lousiness at verbal components seems a bit suspect too.

I just want to mention that if we didn't break down people into racial/ethnic groups for purposes of supposedly "helping" them, we wouldn't have to engage in this kind of nauseating analyses.

[Datepalm]

I just want to mention that if we didn't break down people into racial/ethnic groups for purposes of supposedly "helping" them, we wouldn't have to engage in this kind of nauseating analyses.

If we hadn't broken people down into racial/ethnic groups for the purpose of disenfranchising, exploiting, enslaving and exterminating them, we would have to engage in suppsedly helping them.