Tuesday, March 7, 2017

Academic norms and the Charles Murray incident

I sort of stumbled into the discussion sparked by the incident at Middlebury college via a short tweet storm that ended up getting a lot more traction than usual. It also lead to a discussion with João Eira that I said I would continue in blog format because Twitter is a difficult medium for nuance. There were several economists in my feed that brought the incident up, the issues touch on scientific methodology, and I think I can make a connection to Russ Roberts' piece [7] that has been making the rounds. Surprisingly, there's also a connection to my recently updated comment policy. Therefore I thought it germane to my blog.

So you don't have to go to Twitter, the gist of my tweet storm was that

  1. Pushing bad science on people breaks an institutional norm, so it unsurprising that people reacted by breaking another institutional norm
  2. Charles Murray is a virus of enlightenment values (I'll explain more below)
  3. If Murray was a biologist studying legumes with work of comparable quality, he wouldn't have even been invited to speak (well, unless it was a pro-legume interest group interested in his conclusions)


One thing I'd like to make clear is that I am not encouraging violence. I am saying I understand the strong negative reaction to letting Murray speak at a college that turned violent.


Breaking norms


The normal procedure in science and academic pursuits is to first learn the field of study, then do good quality work, publish in peer reviewed journals, and have those papers cited by either your own generation or younger generations [0]. In short: do useful credible work. Murray has failed to do this with his most broadly known work [1].


Normally if a major piece of your work is as discredited as The Bell Curve, you do not get invited to speak at a lunch seminar, much less at a venue with a broader audience. In this, the people inviting Murray to speak violated academic norms. I tried to come up with a good example of how badly this defies norms but was unable to come up with a real world one. It would be like inviting Martin Fleischmann to speak about cold fusion in a counterfactual world where they didn't retract their paper. Also recognize how ludicrous it is to say Fleischmann must be allowed to speak about cold fusion in the interest of open discussion. Next week, we'll have a speaker tell us 2 + 2 = 5.

In the US, we're starting to get a taste for what happens when someone repeatedly violates institutional norms, but around the world when this happens the result is usually loud protest with the potential for violence. The Rodney King verdict and the subsequent unrest in Los Angeles comes to mind. That verdict was as much of a violation of social justice norms as inviting Murray to speak at a college was a violation of academic norms.

The virus: political norms infecting academic norms

One thing to understand about science is that there is a subtle but important difference in the meaning of "open discussion". In politics, we have the "freedom of speech" norm: people are allowed to say what they want. There is no requirement to have any supporting evidence. There is also no requirement that the speaker be open to another's speech. I can publish garbage if I'd like and I don't have to listen to your criticisms.

Science developed well before the freedom of speech was enshrined in the US Constitution and differs from "free speech": you are required to have evidence and you are required to be open to another speaker's speech. These norms are enshrined in the peer review process of academic journals. If I don't present convincing evidence that my paper is correct or if I don't respond convincingly to the reviewer's questions, my paper doesn't get published. I get shut down. It's an arcane and inefficient process, but it enshrines the norms of academia.

The media is the parallel institution to academic journals that follows political norms. It publishes "both sides" of even scientific issues like global warming, and will print factually false statements by people in the spirit of freedom of speech. Editorializing, allow me to say that an institution that operates this way doesn't seem to serve much purpose unless it calls out lies and factual errors or restricts itself to philosophy.

Murray and those like him are viruses that use these competing processes to propagate themselves. Murray publishes a political tract via the media that looks superficially like science analogous to the way a virus simulates proteins that gain access to the cell's machinery. Open academic discussion begins (e.g. here [pdf]) because the subject has been brought up (i.e. the virus DNA is in the cell). Further data is collected and studies are performed that Murray can selectively cite or perform facile analyses on. This is analogous to using the cell's machinery to produce proteins the virus needs to reproduce itself. Murray however is not open to the refutations of his book. It's never retracted (analogy: programmed cell death). When academic pressure tries to right the wrong by denying him academic positions or not publishing his papers (e.g. the body producing a fever), political pressure says that Murray is being censored, that colleges are not allowing open discussion (and ideological institutions like Mercatus or AEI host him, or request his presence in academic forums i.e. exactly what happened). The subsequent academic discussion of his work enables Murray to publish in academic journals despite the low quality of his research.

Credibility, self-editing, and open dialog

I wrote up a different short tweet storm that dealt with Russ Roberts' views [7] about economic research a few days before that turns out to be related. Roberts suggested a new kind of academic openness where economists publish e.g. all of the different exploratory regressions they tried before arriving at the one they put in the paper. I said that good scientific practice dictates that you should record all of this, but that once the paper is published replication or contrary results (also published) should dictate the debate ‒ "not armchair critical theory analysis of work products" (as I said in the Tweet).

The questioning process of peer review could (and sometimes does) analyze those work products, but the peer review process primarily relies on your academic credibility to allow you to "self-edit" your notes to produce your paper. However, this assumes academic norms, not political norms.

Political norms neither subject you to peer review nor support academic credibility. In short, you cannot trust the self-editing of work produced under political norms which makes Roberts' call to produce the initial work products a reasonable suggestion. One way of re-framing Roberts claim is to say that political norm have infected the academic process in economics and therefore we should give up the traditional peer review process for public review.

I personally like this idea (in fact, I follow it with this blog ‒ I effectively publish my scientific notebook on the internet), but it requires its own norms. One is making everything available (e.g. software, data). That's another one I've followed inspired by Igor Carron and his blog dealing with signal processing and machine learning (here's the hardware and software implementations page). This takes the focus off of peer review and puts it on reproducibility.

Another norm this requires is for the resulting open discussion to be genuinely open in both directions. Reputations need to follow bloggers and blog commenters, and everyone needs to be genuinely interested in dialog and capable of accepting (i.e. being open to) criticism. As I mentioned above, I recently changed my comment policy from one that followed political norms ("free speech") to one that follows more academic norms (if you're not open to being wrong, you're shut down).

No one owns the ideas

Another area where the academic norms and political norms differ is in the treatment of the argument from authority "fallacy" with the latter being a lot more amenable to such arguments. You will see the political norm in action in economics when people appeal to what Keynes or Minsky "really said". 

However the academic norm favors argument from credibility. The name of the person is unimportant. I probably understand quantum mechanics better than Werner Heisenberg ever did. I've built up some academic credibility in quantum field theory by publishing several papers (going through the peer review process) and a doctoral thesis (going through the thesis defense process). I can credibly talk about the ideas of Feynman, Weinberg, or Witten without invoking their names. I lack "authority", but I have credibility.

The flip side of that is that no one owns the ideas. I don't need Feynman's name to support every path integral and I can take the insights of the approach into completely different subjects that Feynman did not foresee. I'm not limited to what Feynman "really said". People are free to take the information equilibrium framework I've been developing on this blog and write their own papers and blog posts.

You may be thinking I've gone far afield from taking about Charles Murray, but this is terribly relevant. If a university prevented him or the odious Milo Yiannopoulos from speaking, this would not be violating of academic norms of openness. It would be upholding the academic norm of credibility. The "ideas" [2] are not being suppressed, the speakers are. If the "ideas" are so good, get someone with academic credibility to expound them in your academic environment. In science, you don't need Einstein [3] in order to talk about general relativity. If the relationship between race and intelligence is such an interesting academic research question surely you don't need famous (but academically discredited) names to talk about it. No one owns the ideas, so someone with academic credibility can talk about them at a university if they really need to be discussed.

In a sense, that gives it away. The odious Milo Yiannopoulos in person was critical to the desires of the young Republicans at Berkeley and the AEI student group needed Charles Murray, not a seminar by one of his acolytes. They needed an argument from authority. It was political, not academic.

That's the difference between political openness and academic openness. Academic openness means you can get someone besides Charles Murray or have him talk at a non-academic venue.

Pushing the ideas of the powerful

The US has supported systemic racism since before it was founded. This is still in place (just read Ta-Nehisi Coates), and therefore it's not like racist ideas don't have a venue or a constituency. Racist ideas are a load of garbage academically, so breaking academic norms in order to push ideology supported by the powerful [4] ... well, rubs a lot of people the wrong way. We also have a president and a party that have broken a lot of norms with the continued support of CEOs and the financial industry. These norms have been broken to push the ideology of the powerful (business interests) or the dominant (white Americans over Syrian refugees). This has resulted in protests. If norms continue to be broken, we can expect those protests to become violent. I'm not advocating violence, but when Republicans spout baldfaced lies and act with rank hypocrisy and there are no consequences to the violations of norms, people will think there are no rules anymore and act accordingly [5].

It is key to understand here that it is power and money behind it. As mentioned above, it seems that academic norms have been routinely violated in economics. Business interests pushed "free markets" not because they were the best theory supported by data surviving rigorous peer review (in fact, that process showed many, many free market failures), but because they served the powerful. CEOs set up pseudo-academic institutions in an effort to infect the academy with the free market virus which spread through "open academic discussion". The global financial crisis sparked a protest ‒ and from what I've read it doesn't discriminate between the good academic economists from the infected ones [6].

If academic and social norms continue to be violated in the interests of the powerful, I fear the protests are only going to get worse.

Footnotes:


[0] Max Planck: A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it.


[1] I am not going to cover the well-documented issues with Murray's "research". Suffice to say that the concepts of "race" and "intelligence" are ill-defined on their own independent of each other, and his work on The Bell Curve was not subject to peer review (or even given to potentially unfriendly reviewers). Additionally, Murray's background is in political science and history so he has limited expertise to speak about intelligence (the domain of neuroscience, biology, or psychology). Overall, The Bell Curve is pseudo-scientific garbage.

[2] I can't put enough quotation marks around the word "ideas" here so I won't even try.

[3] I hope the universe forgives me for putting these names in the same paragraph.

[4] I can hear the precious snowflakes now: "But conservative and racist ideas are an oppressed minority on college campuses." We don't discuss aether anymore, either. It's because those ideas are garbage and you're just a terrible person.

[5] In the language of information equilibrium, they will discover this new previously inaccessible (because of past norms) state space volume and occupy it.

[6] I think a shift from the virus to the zombie analogy is appropriate here as well as John Quiggin's book.

[7] I am referencing Roberts' call to produce work products, not the main point of his article which Noah Smith deals with very well. Updated 11 March 2017.

24 comments:

  1. You do know he wasn't there to talk about The Bell Curve right?

    I also would have liked to see more substantive analysis of why he was wrong rather than just say that he is, but I understand your point better now, thanks.

    Joao.

    ReplyDelete
    Replies
    1. Though I guess there's a lot more non-sense in universities these days, as documented hilariously by RealPeerReview, and they haven't quite been received the way your analysis of how academia works would suggest.

      And I do think you're being political/ideological. Republicans as evil, their ideas discredited, support of the powerful etc etc. I guess you haven't even read a word of Murray because I am pretty sure you're almost calling him racist but a) never seen him saying racist stuff and b) quite the contrary. His letter to Virginia Tech might be of help here:

      - https://www.aei.org/publication/an-open-letter-to-the-virginia-tech-community/

      Delete
    2. I understand he wasn't talking about The Bell Curve, but as I mention above The Bell Curve was the reason for his loss of academic credibility. Until he disavows The Bell Curve he has not regained that credibility.

      Re: RealPeerReview

      I followed that on Twitter for awhile until it seemed to just be anti-gender studies and just stating the papers must be obviously wrong.

      Re: ideology

      That's always the issue when you say someone is being ideological: if I write something against an ideology, then I must be ideological with the anti-ideology mirror-image.

      Unfortunately were that argument valid it would work for any criticism of any ideology. I'm also not sure if you have listened to Republicans any time recently.

      Re: racism

      I did in fact read The Bell Curve in college where we discussed its views on race. When Murray says the book is not about race in your AEI link, that is a baldfaced lie. All one has to do is look at the summary available at the AEI itself:

      https://www.aei.org/publication/the-bell-curve/

      ... Part 2: Cognitive Classes and Social Behavior Poverty Schooling Unemployment, Idleness ... Welfare Dependency Parenting Crime Civility and Citizenship Part 3: The National Context Ethnic Differences in Cognitive Ability Ethnic Inequalities in Relation to IQ The Demography of Intelligence ... Part 4: Living Together ... Affirmative Action in Higher Education Affirmative Action in the Workplace ... Appendix 7: The Evolution of Affirmative Action in the Workplace

      These terms are all very loaded racial terms in US political discussion. Obviously the AEI thinks it's about race.

      Let me also let you know that I grew up in a conservative family in Texas. I've lived among it. I know the culture. This is what racism sounds like. Until recently, most "respectable" racists didn't say overt racial slurs but instead refer to "those people" and use adjectives like "ethnic" or "idle". As my English teacher used to say: it's the connotation, not the denotation.

      But you could even read Murray's open letter. He starts saying the book is not about race, but then has a whole section where he reiterates the books conclusions about race saying they haven't been refuted. That's some strange cognitive dissonance.

      Delete
    3. You've given me a lot to think about Jason, even though I am still not convinced by your approach to arguing your case, especially calling Murray a racist, or implying he is one.

      My ideology comment wasn't about defending Republicans (come on now) but the way it seems to me you are approaching this issue: not intellectually, or dispassionately, but in a sort of combative way because some bad people have used Murray's work to speak insanity. That was my whole point about this topic, if you're going to argue your case about why Murray's ideas are like a virus then actually make a case for it..

      I guess what I am getting at is that you could have convinced me further if you were to have at least tried to pass the Ideological Turing Test then make a case against it.

      This has gotten long already so I'll refrain from commenting further on the topic. Also, because sometimes the internet makes these things look like a bloodsport, I want to say your blog and your ideas are some of the most interesting things I have found recently (though I wish you published some code that wasn't Mathematica so I could actually take a look at it :P )

      Hope all is well,
      Joao

      Btw: Dalton Conley has a recent paper on The Bell Curve if you're interested. I haven't read it yet, but you can find it here: https://www.sociologicalscience.com/articles-v3-23-520/

      Delete
  2. Pretty amazing post Jason! Shorter Symbyiotic: I don't care if Murray's been proven to be a racist scientific fraud, I still want him to pretend to have something important to say because he's on my political team.

    ReplyDelete
  3. This post offers the clearest perspective on this I've seen. Well done.

    ReplyDelete
  4. That's extremely unfair. Symbyotic said nothing about identifying as a Republican, merely that it was wrong to stereotype Republicans as evil, racist, supportive of the powerful, etc.

    I'm very distant from Murray with regard to political views, but I found much to like in the Bell Curve. It's support for gifted education is admirable and I found very little in the book that would justify treating people of different ethnicities unfairly; Murray and Hernnstein go out of their way to denounce anyone who would use their work for racist ends.

    Granted, its possible to criticize the bell curve on its statistics. The authors used methods with a less than ideal level of rigor (they claim that the book was meant to "start a conversation which other researchers could continue with more sophisticated statistical methods", and claimed that using anything other than the most basic of statistical techniques could lead their intellectual opponents to "unfairly claim they were skewing the data". Whether you think these excuses are legitimate is up to you.)

    Furthermore, if anyone is discredited here it is Hernnstein, not Murray. As Jason says, Murray has a background in political science; most of the Bell Curve's actual content came from Hernnstein, who had a psychology background. Murray was solely there to formulate the policy implications of Hernnstein's research.

    On the subject of Bell Curve's content, I've also read several papers from the APA on psychometrics. The Bell Curve's claims don't seem all that far away from the norm. If you're left-leaning and want to see some of the same research discussed from a different political perspective, I highly recommend James Flynn; the Flynn effect is named after him, and he's pretty respected among his colleagues.

    Maybe I'm wrong, and the Bell Curve is actually a giant discredited mess. But given that I've taken the time to familiarize myself with Murray's work and didn't find the critiques leveled at it very convincing, I'd say it is at least *possible* for a well-intentioned, reasonable person to come away from the Bell Curve with the same perceptions as I did. You can't claim that Symbyotic has political motivations; its at least epistemically charitable to call him *mistaken*, and explain why he's wrong.

    Which I'm open to other people doing for me, of course. I'm genuinely curious why people find the Bell Curve discredited--what parts of it are discredited, and what sources did the discrediting? I ask not because I want to start a giant internet flame war, but because I'm don't want to support of a work that's actually incorrect and would really appreciate to be led to resources showing what I'm missing.

    ReplyDelete
    Replies
    1. Sorry, this was meant to be a reply to Todd Zorick.

      Delete
    2. Wow this post really brought out an interesting crowd.
      SilasLock: I found much to like in the Bell Curve

      2 minutes on the Google led me to very cogent article that showed that even in 1994 the obvious problems with the data in the Bell Curve cast doubts on any conclusions it came up with, regardless of the eugenicist overtones. Charles Lane's 1994 takedown shows that much of the data was cherry-picked from shoddy articles at a white supremacist pseudo-scientific journal founded by a South African pro-apartheid Boer. Now you seem like a smart guy. There are a couple of possible reasons why you might have missed this in your "research". 1: You are actually on the wrong part of the Bell Curve yourself; or 2. you happened to be lazy in your understanding because it supported your Bayesian prior. I'll leave it to you to figure out which is more likely.

      Delete
    3. SilasLock, you can refer to my comment above Re: racism. Most of the time I hear anyone defending so-and-so as not racist it makes me think of the scene in The Shawshank Redemption where Andy asks the warden: "How can you be so obtuse?"

      White people may beat other ethnicities on IQ tests, but they certainly would fail a basic reading comprehension test asking whether a particular piece of writing was racist.

      But as I mention below, single-factor (g-factor) intelligence correlates with various cognitive abilities at say a correlation coefficient of 0.4-0.5. That is used as evidence in favor of g-factor theories. However, g-factor correlates with socio-economic status at also about the 0.4-0.5 level. It seems odd to promote one set of 0.5 correlations and demote another.

      And in any case without a second Earth that didn't have systemic racism and oppression, any extraction of the counterfactual would be strongly model dependent.

      When you say it's Hernnstein and not Murray that's pretty laughable. Are you saying Murray was duped? But then you are also saying the conclusions are supportable. And Murray himself seems to defend the actual content in Symbiotic's link above.

      As Todd mentions, confirmation bias is a hell of a drug. But there's one thought experiment I like to suggest to people who try to defend racist views as not being racist.

      Close your eyes, take a deep breath. Now think of a black person. How do you feel about that black person? You're the only person who knows how you really feel, so no one will ever be able to use this as evidence against you. Maybe it's not racism. Are you fooling yourself? Because you're not fooling anyone else.

      Delete
  5. Mr. Smith I think is mistaken on IQ and race. Yes, cognitive ability is a real thing. Yes, average cognitive ability differs between ethnic groups. And finally, most experts in the field actually do attribute these differences in part to genes. IQ is highly heritable, and the distribution of alleles contributing to cognitive ability are not uniformly distributed across populations. Smith's reaction expresses a moral sentiment; it represents the prevailing social norms of our era, and I think is indicative of the powerful social taboos against recognizing average biological group differences. To my knowledge, the one chapter in the Bell Curve examining ethnic differences in IQ has not at all been discredited. Interestingly, Gould actually did commit academic fraud, but was never publicly excoriated over it. I realize this isn't the focus of this particular blog. Zealots are not going to be persuaded, and most everyone else can feign disinterest or be satisfied in parroting worn out, sophistic arguments like 'race doesn't exist', Lewontin's fallacy, or suddenly become an obscurantist over things like analysis of variance, etc. For the few newbies who are genuinely interested in this topic, a decent primer citing the peer-reviewed literature can be found here:
    http://thealternativehypothesis.org/index.php/2017/01/07/race-and-iq-the-case-for-genes/

    I also agree that Flynn is probably the most well-known environmentalist in the field, although he is in the minority.

    ReplyDelete
    Replies
    1. 1. It's Dr. Smith.

      2. Single- and two-factor theories of intelligence show that the correlation with socio-economic status is as strong as the single factor correlation among cognitive abilities.

      3. I assume you have discovered a second Earth where there wasn't 500 years of enslavement, racism, and oppression whereby we can extract the counterfactual?

      4. Race doesn't exist as a scientific concept (biologically, it would be phenotype which also includes things like the varying fur of domestic cats). It does exist as a social concept and in that domain it's very important.

      5. OMG that website is super racist.

      Delete
    2. Re: #1. Certainly, Dr. Smith. My apologies. I wasn’t expecting to get a reply to my post, but since you actually make a couple arguments, I’ll use this occasion to present the standard hereditarian position. My goal is to convince you (or a reader) that the hereditarian argument for racial IQ differences is at least as strong as the environmentalist position.

      Let me first speak broadly about motives. The Lane piece cited above is a good example of how this topic is usually addressed in the media. There is no honest assessment of evidence to evaluate competing hypotheses. There is a general consensus that ethnic differences in tests of cognitive ability and other standardized tests exist; are persistent over decades and in various countries; are generally reliable; and culturally unbiased. The few 'environmentalist' hypotheses that could plausibly account for some of the differences in standardized test performance are NOT the ones usually cited in popular media. The gaps in test performance are almost certainly NOT caused by poverty, discrimination, or secret white-supremacy-mind-rays that somehow also boost Asian performance, are invariant to the actual presence of whites; are to a large extent invariant to family upbringing; and magically produce the same effects upon culturally diverse peoples all around the world, but only to the extent that they share common ancestry. Over the decades one ad hoc theory after another has been proposed, tested, and nearly all have been largely dis-confirmed.

      Biological denialists usually adopt two approaches: first, they engage in name calling, and second, they point to the mere possibility that environmental factors could explain racial IQ gaps without actually providing evidence that environmental factors actually do explain the gap. Moreover, because of the ‘sociologists’ fallacy’ (i.e. genetic confounding), to prove their case environmentalists actually must go one step further and show that environmental factors themselves not correlated with genes explain racial IQ gaps. This hasn’t been done. Therefore, the weight of the evidence right now is that the gap has a partial genetic cause.

      Murray is the exception and not the rule - even that damn chapter in that book 20 years ago was the exception. It was only one chapter! In reality, the topic of biological group differences is taboo. The student protesters are not speaking truth to power. They are the power, at least one wing of the establishment enforcing ideological conformity.

      You seem to think that the hereditarian position could only possibly be motivated by a prejudice, a desire to want to believe it, perhaps because it is comforting to believe. The truth is the opposite. Most whites I’ve ever encountered think exactly as you do. There is no psychological comfort in holding this view. It is horrifying and difficult and much easier to avoid thinking about seriously. Plus, of course, espousing hereditarian views of racial IQ differences immediately elicits accusations of ‘racism’ (i.e. of being a very bad and immoral person). No argument whatsoever could ever convince most people that biological group differences even exist.

      This idea that hereditarians (at least the professional ones) are somehow in it for the money, or the ego, or whatever - is crazy. It’s much much easier to ignore this entirely. Social, personal, and financial incentives all militate strongly against adopting the hereditarian viewpoint, even though, I would argue, it is the one that on some level has the most prima facie plausibility. Denying group biological differences been a dogma in academia now for decades, but hopefully this is changing somewhat. See, for instance, this paper: Human Biological and_Psychological Diversity

      Delete
    3. Re: #2. The implication of this fact isn’t clear to me. ‘G’ correlates with SES, as do the results of intelligence tests generally. The relationship between intelligence and SES would I think be expected by hereditarians and environmentalists. So the mere existence of the correlation itself doesn’t bear upon the causal direction of the two, the within-group heritability of either, or upon the heritability of between-race IQ differences.

      In addition, I should point out that whether or not we choose to represent intelligence as a single general factor or several is actually not relevant to the interpretation and possible genetic causation of racial differences in intelligence. Factor analysis is just a method of representing data. It is true (and surprising) that the IQ subtests correlate with one another. To take a common standardized test such as the SAT, people who have above-average verbal scores also tend to have above average math scores. But so long as intelligence tests are unbiased, the races must still differ in whatever it is that intelligence tests measure. It doesn’t matter to me whether we call this a gap in general intelligence (g) or a gap in reading and math, or as (possibly heterogeneous) differences in the primaries.

      Having said that, however, the correlation of the first principle component of an IQ battery with SES is actually a good reason to think that this factor exists (or is useful as a descriptive device), i.e. is capturing to some degree learning ability and is not just a mere artifact. Speaking generally, a sufficient condition for the reality of any posit it its correlation with a wide range of independent variables. Something must be generating the correlations. This is why a thermometer reading is real, because it has real-world correlates. In contrast, consider something like AQ - a person’s height divided by his street address. Although clearly defined, nobody thinks it corresponds to anything because it doesn’t correlate with anything real; it lacks predictive power. The correlates of IQ are evidence of its reliability and predictive validity.

      SES and IQ correlate; and SES itself is highly heritable - i.e. can be attributed to genetic variance. According to a twin study by Krapohl et al 2014, intelligence was (unsurprisingly) the single most important hereditary trait determining educational attainment, but that other genetically influenced traits were also important. A genome-wide association study by Rietveld et al 2013 also found multiple ‘Genetic Variants Associated with Educational Attainment’. Here’s a more recent study by Okbay et. al. 2016 published in Nature identifying 74 loci associated with educational attainment. These studies look primarily at educational attainment but there are also strong correlations between IQ and income, occupational status, job performance, crime (inversely related), etc. More pertinent to the issue of SES and IQ is to what extent parental SES influences adult IQ. Strenze 2007 shows that intelligence is a “better predictor than parental SES or grades”, although it was not an ‘overwhelmingly’ better predictor. I think it is reasonable to suggest, and is probably the consensus view, that SES and IQ both influence each other to some extent, and both are heritable. At any rate, I’m not sure what you’re getting at and this has no bearing on racial IQ differences, at least not directly.

      Delete
    4. Re: #3. The hereditarian position is that genotypic variance explains part of the between-race phenotypic variance in intelligence. Environmentalists argue that genotypic variance explains NONE of the between-race phenotypic variance in intelligence. I assume you defend the latter view. That IQ is heritable within groups is to my knowledge not challenged by anyone. The basic findings over hundreds of studies is that IQ is largely heritable - genetic variation explains/accounts for about 70-80% IQ variance; shared (i.e. family) environment accounts for about 20% of IQ variance; and unshared environments accounts for the remainder and matters the least. In addition, the heritability of IQ tends to increase as we age (the Wilson Effect), reaching estimates of around .85 for the adult population. Traditionally these estimates have been based on studies of monozygotic twins, dyzogotic twins, siblings, etc. but more recently ‘Genome Wide Association Studies’ (GWAS) now enable scientists to map entire genomes, thus effectively controlling for differences in DNA directly.

      The obvious implication is that differences in group averages are also to some extent heritable - i.e. ‘due to genes’ This of course does not *necessarily* follow, although within-group heritability does make between-group heritability more likely in my view. There are two possible environmentalist explanations of the Racial IQ gap: 1) explanations that invoke environmental factors that also vary within groups (i.e. that also factor into within-group variance); and 2) explanations that postulate a factor that has no within-group variance but which is consistently present in one group and consistently absent in the other group. Jensen calls the latter ‘X-factor theories’. I would say that SES fits into the first category. The slavery and oppression unique to African blacks who were enslaved and oppressed belongs to the second ‘x-factor’ type.

      Regarding the first type, e.g. SES, for between-group difference to be fully explained by SES (or other varying environments that also account for within-group variance), the two groups have to differ to a very large extent. The average IQ gap between whites and American blacks is about 1 (white) standard deviation, or 15 pts. Assuming a within-group heritability of .5, mathematically this would mean that the black-white average difference in SES would have to be at least 1.41 standard deviations (of the distribution of the environmental component of IQ, e.g. SES). If the heritability of IQ is .6, the required SES mean difference would be 1.58 SD's; for a heritability of .7, 1.82, and for a heritability of .8, a difference of 2.24 SD's. For a nice chart, see (Sesardic, Making Sense of Heritability, pg. 139). The higher the heritability, the higher the burden for the environmental theories to explain the gap.

      The second type of argument, the x-factor argument, at first seems convincing. Blacks, as a group, have been at some times and places exposed to a kind of pervasive discrimination triggered by their phenotype not faced by whites. Could this plausibly explain the the pervasive black-white IQ gap? Flynn, the leading environmentalist in the field, doesn’t think so. He wrote that this was:
      “simply an escape from hard thinking and hard research. Racism is not some magic force that operates without a chain of causality. Racism harms people because of its effects and when we list those effects, lack of confidence, low self-image, emasculation of the male, the welfare mother home, poverty, it seems absurd to claim that any one of them does not vary significantly within both black and white America. (Flynn 1980: 60)

      In other words, the so-called x-factor, e.g. discrimination, would have to operate locally through similar mechanisms involving factors that vary within groups. Then these hypothetical mechanisms would be faced with the same implausible explanatory burdens as SES, single motherhood, self-esteem, and so on, which seem incapable of explaining the gap entirely.


      Delete
    5. Regarding your point about slavery and oppression, we don’t need a parallel universe, we just need variance. Below I present some of the most salient findings in this line of research, presented mostly as ‘stylized facts.’ The list isn’t exhaustive, but I think examining them in toto renders the (extreme) environmentalist position less plausible than the hereditarian alternative.

      1. Why do groups who share common ancestors tend to have more similar IQs, regardless of where they live currently, whether they are minority or a majority, regardless of the political history, type of government, or standard of living of the country in which they live? For example, East Asians have higher IQ’s than blacks, basically everywhere, on average. Why would black Canadians and black Swedes of say, West African origin, have more similar average IQs to one another than to white Canadians or native Swedes?

      2. The rank order of standardized tests is the same in virtually every county and school district: Asian > White > Hispanic > Black. Why is this pattern virtually constant across thousands of districts? Why aren’t there more exceptions? If this were cultural, why would this not vary more widely than it does?

      3. If discrimination causes all of racial IQ differences, why would the IQ gap be relatively stable over the course of a century?

      4. Why does the gap appear to be invariant to the actual presence of whites? Average black IQ is roughly similar across cities, counties and states regardless of how many or what % of whites live there. This is also true for countries. European admixture raises IQ, but the presence or absence of whites does not, at least not much.

      5. Why would mix-raced children (of black and white parents) have an average IQ in-between whites and blacks? Colorism is a possible explanation. But I think this result holds even for mixed-race people who didn’t know they were mixed race.

      6. Controlling for SES, and/or income alone does not eliminate the black-white IQ gap. Rich blacks perform worse on standardized tests than poor whites. (I've verified this myself analyzing the AFQT data; it's also true for SAT, ACT, GRE, and other tests) It seems counter-intuitive that rich blacks are still oppressed more than poor whites.

      7. Why would the gap exist even among adoptees adopted into white homes? (e.g. the Minnesota Twin Study results). Perhaps they were treated differently in the school system which differentially impacted their cognitively ability...

      8. One possible environmentalist hypothesis is that racism depresses IQ via the mechanism of low self-esteem. The problem with this is that the data refute it: blacks actually have the highest average self-esteem. (Twenge and Crocker, 2002).

      9. Stereotype threat (i.e. test anxiety arising because blacks or women are made aware of being black or a woman, which is supposed to affect all blacks and women to a great enough extent to matter) has also by now been thoroughly debunked in recent meta-studies (just do a quick google search).

      Moreover, there seem to be biological predictors of IQ that also differ between races which are not explained by the environmentalist hypothesis.

      10. Reaction time correlates with g, and differs from mere muscle speed. Although blacks have faster average muscle speed, they have slower reaction times on average than whites and Asians.

      11. What is the hereditarian explanation for the fact that the most g-loaded components of IQ tests are also the more heritable?

      12. Why do IQ gaps appear in 3 year olds? Obviously education gaps can’t explain this, although other SES factors might.

      13. Finally, why would would average brain sizes differ among racial groups and also correlate with IQ? Yes, this is actually a thing, and is actually real. Gould was wrong. It turns out that Cyril Burt's work has largely been vindicated, and that a correlation does exist between brain size and intelligence.

      Delete
    6. My LAST POST.

      Re: #4: RACE. The issue of ‘race’ to me is a red-herring. Nothing changes if we drop the term and use instead deme, ethnicity, ancestry, or genetic clusters. Phenotypically, races are identified on the basis of clusters of traits, reflecting common ancestry. Genetically, races can be identified and distinguished not (usually) on the basis of the presence of absence of single alleles, but on the basis of different frequencies of alleles.

      Colloquially, as well as scientifically, race just refers to birthplace of ancestors. There is some difference of opinion regarding whether genetic variation is entirely "clinal" (gradual, continuous) or whether it is useful to refer to geographical discontinuities, 'clusters' or categorical 'classes' as well. But this has no bearing on the white-black IQ gap. We don’t need an ‘essentialist’ understanding of race as one of discrete genetic categories to entertain the possibility that the frequency distribution of alleles contributing to intelligence differs between populations derived from sub-Saharan African populations and those derived from continental Europe.

      Some argue that racial categories are arbitrary, for example, that the algorithms that identify human groupings based on genetic clusters simply reinforce pre-existing racial categories. There is an interesting discussion of this in a recent volume in the journal Sociological Theory, “The Genomic Challenge to the Social Construction of Race”. (2014). Shiao et al. explicitly discuss this, saying that the latest algorithms don't detect genetic similarity on the basis of pre-defined groups. The largest clusters that can be identified correspond to the five continental races: sub-Saharan Africans, Indo-Europeans, East Asians, Australian aborigines, Native Americans. If we cluster more finely, the people of the Indian subcontintent and the people of the Middle East constitute a 6th and 7th grouping.

      My notion of race is basically what Hoschild (et al.) in Creating a New Racial Order - a book very friendly to social constructionist arguments - mean by the term ‘statistical’ race (as opposed to typological concepts of race), according to which, "people of common ancestry are more likely to share certain genes or alleles (versions of genes) than those who do not share ancestry." Put simply, on average, members of a race are more closely related genetically to one another than they are members of other races. (see Shiao 2014: p. 250)

      One often-repeated argument is that races can’t matter because humans share 99% of our DNA. The conclusion doesn’t follow from the premise, however, since the remaining 1% can matter quite a lot. After all, we also share 95-98% of our DNA with chimps and gorillas.

      Lewontin famously found that ~85% of total human genetic variation occurs within a population; ~8% occurs between two populations of a given race; and only ~6% of variation accounts for variance between races. From this he concluded that races were taxonomically meaningless. This has been criticized as ‘Lewontin’s Fallacy’ - the ‘fallacy’ being the assumption that genetic differences between populations are uncorrelated with one another; if they are correlated, they become much more significant. That remaining 15% matters. In other species, biologists typically regard the degree of genetic differentiation seen between the continental races as evidence of sub-species.

      Moreover, that there is more genetic variation within groups than between is true because humans are members of the same species. Average group differences can still occur and be important. The variance of height within sexes is greater than the variance between but men are still on average taller than women.

      I also readily acknowledge and concede the following: that races change over time via inter-breeding and migration; that admixture occurs (e.g. mestizos); that social perceptions of ancestry have changed; and that racial classification schemes have changed over time.

      Delete
    7. You cited an incredibly racist/alt-right website above ("The Alternative Hypothesis"), so your academic credibility is pretty low on this subject.

      Per my post above, without credibility we cannot trust you to self-edit your prose or that your citations constitute "Feynman integrity": bending over backwards in exploring the literature for contradictory information or considering your conclusions may well be wrong.

      In a normal academic process, your "paper" would be rejected in peer review based on citing discredited sources and no one would ever see it. The analog for comments on this blog is that they'd be deleted.

      However I'm leaving your posts up as an example of what I mean by a virus mimicking cell receptors (pseudo-"open academic discussion") attempting to co-opt the cell's machinery into continuing the virus life cycle (producing more pseudo-"open academic discussion").

      It's also an example of the specific lack of self-awareness on display. Charles Murray and websites like "The Alternative Hypothesis" (check the "about us" section, but I'm not going to link) are aware that people perceive what they are saying is racist, but are completely unaware of which specific behaviors they exhibit that lead to that perception (e.g. not citing contrary evidence/results).

      Delete
  6. O/T: Jason, imagine you could select people to be in a new nation that you would inhabit as well. Now imagine the most right wing and most left wing people in the population. What persons (past or present) exemplifies the most rightwing acceptable camp and the most leftwing acceptable camp? For example, maybe John Podherotz exemplifies the most right wing acceptable person, and Bernie Sanders the most left wing acceptable person. Who would you choose?

    ReplyDelete
    Replies
    1. Not sure I can answer that with a single left-right dimension. I am willing to tolerate much broader ideological disagreement on e.g. monetary policy than e.g. racism.

      I can say that while I disagree with him on almost everything, John Cochrane is at least challenging in a good way. And Bernie Sanders represents the far edge of "his heart is in the right place" tolerance for making claims that don't stand up to rigorous analysis. That said, I'm open to far more 'socialism' than even Sanders was advocating.

      Delete
    2. Thanks for the reply. Good article too BTW.

      Delete

Comments are welcome. Please see the Moderation and comment policy.

Also, try to avoid the use of dollar signs as they interfere with my setup of mathjax. I left it set up that way because I think this is funny for an economics blog. You can use € or £ instead.