[Epistemic status: uncertain. Everything in here seems right, but I haven’t heard other people/experts in the field talk about this nearly as much as I would expect them to if it were true. Obviously amount of variability attributable to environment (shared and non-shared) increases as the variability in environments in the sample increases]
The “nature vs. nurture” question is frequently investigated by twin studies, which separate interpersonal variation into three baskets: heritable, shared environmental, and non-shared environmental. Heritable mostly means genes. Shared environmental means anything that two twins have in common – usually parents, siblings, household, and neighborhood. Non-shared environmental is everything else.
At least in relatively homogeneous samples (eg not split among the very rich and the very poor) studies of many different traits tend to find that ~50% of the variation is heritable and ~50% is due to non-shared environment, with the contribution of shared environment usually lower and often negligible. This is typically summarized as “50% nature, 50% nurture”. That summary is wrong.
I mean, it’s tempting. All these social developmentalists were so sure that the way your parents praised you or didn’t praise you, or spanked you or didn’t spank you, had long-lasting repercussions that totally shaped your adult personality. The underwhelming performance of shared environment in twin studies torpedoed that whole area of study. But at least (these scholars of social behavior could tell themselves) it provided a consolation prize. The non-shared environment contributes 50% of variation, just as much as genes. That means things like your friends, your schoolteachers, and even that time you and your twin got sent away to separate camps must be really important. More than enough there to continue worrying about how society is Ruining The Children, right?
Not necessarily. Non-shared environment isn’t really “non-shared environment” the way you would think. It’s more of a dumpster. Anything that isn’t genetic or family-related gets tossed into the non-shared environment term. Here are some of the things that go into that 50% non-shared environment:
1. Error. Measurement error is neither genetics nor family, so it ends up in the non-shared environmental term. Suppose you’re studying intelligence, and you make a bunch of twins take IQ tests. IQ tests measure intelligence, but not perfectly. For example, someone who makes a lucky guess on a multiple choice IQ test will get a higher score even though they are not more intelligent than someone who makes an unlucky guess. Someone who takes the test when they’re tired and stressed may get a lower score even though they’re no less intelligent than somebody else who takes it well-rested and feeling good.
Imagine a world where intelligence is entirely genetic. Two identical twins take an IQ test, one makes some lucky guesses, the other is tired, and they end up with a score difference of 5 points. Then some random unrelated people take the test and they get the 5 point difference plus an extra 20 point difference from genuinely having different IQs. In this world, scientists might conclude that about 80% of IQ is genetic and 20% is environmental. But in fact in terms of real, stable IQ differences, 100% would be genetic and 0% environmental.
This gets even harder when trying to measure fuzzier constructs like criminality. Suppose someone does a twin study on criminality and their outcome is whether a twin was ever convicted of a felony. This depends partly on whether the twin is actually the sort of person with criminal tendencies – but also partly on whether a policeman happened to be in the area to catch them, whether their lawyer happened to be good enough to get them off, whether their judge was feeling merciful that day, et cetera. Imagine a world where criminality is entirely genetic. Identical Twin A becomes a small-time cocaine dealer in a back alley in West Philly, sells to an undercover cop, and ends up in jail. Identical Twin B becomes a small-time cocaine dealer in a back alley in East Philly, doesn’t run into any undercover cops, and so avoids conviction. This shows up as “variation in criminality is due to non-shared environment”.
Riemann and Kandler (h/t JayMan) run a study which is an excellent demonstration of this. Classical twin studies sometimes use self-report to determine personality – ie they ask people to rate how extraverted/conscientious/whatever they are. These studies find that most personality traits are about 40% genetic, 60% non-shared environmental. Riemann and Kandler obsessively collect every possible measurement of personality – self-report, other-report, multiple different tests – and average them out to get an unusually accurate and low-noise estimate of the personality of the twins in their study. They find that variation in personality is about 85% genetic, 15% non-shared environmental. So it looks like much of the non-shared environmental variation in traditional studies of personality was just error.
2. Luck of the draw. Bob becomes a junior advertising executive at Coca-Cola, where he designs a new ad targeting young female consumers. His identical twin Rob becomes a junior advertising executive at Pepsi-Cola, where he designs his own new ad targeting young female consumers. Both ads are very successful – in fact, exactly equally successful. But Coke’s CEO is a crony capitalist who wants to replace everyone in the company with his college buddies, so he ignores Bob’s good work and demotes him to a low-level position. Pepsi’s CEO is a skilled leader who recognizes good talent when she sees it, and she promotes Rob to Vice-President Of Advertising.
Now a scientist comes along, does a twin study on them, and finds that they have very different levels of income. She reports that there’s a lot of difference between these two identical twins, so much of income must be non-shared environmental.
Science reporters read the study finding that much of the variation in income is non-shared environmental, and conclude that despite their identical genes, there must be deep and mysterious differences in Bob and Rob’s abilities and business acumen. They speculate that Rob had a very inspirational teacher in school who pushed him to achieve greatness, and Bob must have fallen in with a bad peer group who didn’t value hard work.
But actually, Bob and Rob are completely identical in every way, no incident in their past did anything to separate them, and Bob just ended up working for a crappy CEO. In this scenario, inherent predisposition to earning money is exactly the same in both twins, they just have different amounts of luck at it. If both twins become pathological gamblers, but one of them hits the jackpot and the other goes broke, that will show up as “non-shared environment” too.
3a. Biological random noise. The genome can’t encode the location of every cell in the body. Instead, it specifies high-level processes which create lower-level processes which create those cells. But this gives the lower-level processes a lot of leeway, meaning that there can be significant biological differences between identical twins.
Consider by analogy The Postmodernism Generator. It’s a cute program that will make a (sort of) convincing sounding postmodernist essay on demand. We can imagine hundreds of different programmers all designing their own postmodernism generators. Some would be really brilliantly designed and consistently come up with plausible looking essays. Others would be poorly designed and consistently come up with crappy essays that don’t convince anybody. But there would also be variation within the results of each generator. There might be a generator that is mostly terrible but occasionally by coincidence comes up with a really funny essay, or vice versa. In this analogy, the genes are the code for the generator, and the person is an individual essay produced by that generator.
Thus, identical twins have different fingerprints, different freckles, and different birthmarks. Only about a fifth of left-handers’ identical twins will also be left-handed. And twins even look different enough that their friends and parents eventually learn to tell them apart. All of these are non-genetic issues likely to show up in “non-shared environment” but not related to schools or peers or “nurture” as traditionally conceived.
3b. The immune system. Immunology is still poorly understood, but it seems very important. Immune reactions and neuroinflammation have been implicated to one degree or another in a lot of psychiatric diseases. A functional immune system can protect good health; a dysfunctional immune system can make someone constantly tired and miserable.
There seems to be more of an element of chance to the immune system than to a lot of other bodily processes. Part of it is the input – one child in a twin pair might inhale a particle of cat dander at a critical time; another might get some unknown adenovirus with no immediate effects but which contributes to obesity twenty years later. Another part is the output; sometimes a natural killer cell stumbles across something quickly and takes it out without any fuss; other times the immune system misses it for a while and it gets more of a chance to spread.
The end result is that immune-system-related-conditions are really discordant across identical twins. If your identical twin has asthma, there’s only a 33% chance you’ll have it as well. If your identical twin has Crohn’s disease, a disabling autoimmune intestinal condition, there’s only a 50% chance you’ll suffer the same. I’m not sure how significant this is in the broad scheme of things, but I suspect more so than people think.
3c. Epigenetics We know that identical twins have substantially different epigenetics, and there are hints that this underlies discordant behavior. This is probably really important, but I feel bad bringing it up because it seems to be passing the buck. We usually think of epigenetic differences as a response to different environments or life choices. But if identical twins start with the same environment and can be expected to make the same life choices, why do they end up with different epigenomes? I’m not sure what to do with this one.
3d. Genes that differ between identical twins. Apparently this happens! Identical twins come from the same zygote, which means they start out with the same genes, but after that all bets are off. If there’s a mutation in one twin in the first embryonic division, then half of that twin’s cells will carry that mutation. Remember that there are a lot of divisions and opportunities for mutation before any cells even start forming the brain, and any mutation before that time could be transmitted to all brain cells. One study found that the average identical twin pair probably has about 359 genetic differences occuring early in development.
I’ve grouped 3a, 3b, 3c, and 3d together as possible biological sources of variation. One of these – or maybe some 3e I don’t know about – is probably the reason for less-than-perfect twin concordance in conditions like Parkinson’s disease, migraine, autism, and schizophrenia. Needless to say, anything that can make you schizophrenic can probably affect your personality and life outcomes pretty intensely.
But all of this gets counted as “non-shared environment” in a twin study, and used to play up the importance of schools and peer groups.
4. Actual nurture. Twins do have different experiences growing up. How much does this shape their adult traits? Can we separate this out in to specific experiences that shape adult traits, like school and summer camp?
The good news is that Eric Turkheimer has a big review article on this; the bad news is that the discussion section is called “The Gloomy Prospect” and says:
Quantitative analysis of studies of specific nonshared environmental events shows that effect sizes measuring the effects of such variables on child outcomes are generally very small. Effect sizes are largest when confounds with genetic variability and outcome-to-environment causal effects are not controlled. When such confounds are controlled, as in the most recent reports from the NEAD project, effect sizes become smaller still.
I’m not sure if this table represents the “very small” uncontrolled or the “smaller still” controlled sizes
The paper concludes: “We emphasize that these findings should not lead the reader to conclude that the nonshared environment is not as important as had been thought.”
But although I have a huge amount of respect for Turkheimer, I kind of want to conclude that the nonshared environment is not as important as had been thought. My guess is that the nonshared environment as Turkheimer discusses it – differential parenting, schools, peers, and so on – is only a fraction of the “nonshared environmental” term in genetics studies.
If that were true, it would mean that nature is more important than we thought relative to environment in terms of things we can understand and possibly affect. That would make the quest to change important outcomes like intelligence, personality, income, or criminality by changing society even more daunting. And it would make the opportunity to change those outcomes through genetic engineering even more tempting.
IQ tests measure intelligence, but not perfectly. For example, someone who makes a lucky guess on a multiple choice IQ test will get a higher score even though they are not more intelligent than someone who makes an unlucky guess. Someone who takes the test when they’re tired and stressed may get a lower score even though they’re no less intelligent than somebody else who takes it well-rested and feeling good.
Also, experience with the test. Retaking the SAT was generally worth about 150-200 points out of 1600, just because people better used their time and weren’t thrown by the questions.
Classical twin studies use self-report to determine personality
Isn’t self-reporting thought rather poorly of, especially for things that aren’t extremely concrete?
Are you sure? According to here retaking the SAT is worth 10-20 points out of 1600
But those taking the exam five times improved their score by 83 points out of 1600 between the first and last time (see Table 2).
Having taken the SAT, I would feel no surprise concluding that
(A) There’s at least 83 points worth of very studyable/practice-able material in the exam,
but I’ll rule out other explanations first.
Maybe
(B) People get (anywhere close to) 83 SAT-points smarter between Junior and Senior years.
However, looking at Table 6, the time gap between tests doesn’t matter (with the exception that those with only a summer vacation between tests improve slightly less), and people retaking only once show much smaller gains.
Of course there’s also
(C) Students retake tests because they screwed up the previous time(s),
but this effect should be negligible because the mean improvement between “1st time” and “2nd time” looks independent of the number of tests eventually taken.
So it looks like those 83 points come mostly from practice with the SAT and studying. Anything I’m missing?
(Note: I’m writing this primarily for people considering taking the test again.)
The scoring also changes from test to test based on an equating formula. Collegeboard claims that it “ensures that the different forms of the test or the level of ability of the students with whom you are tested do not affect your score”.
https://collegereadiness.collegeboard.org/sat-scoring-before-march-2016
During one test you may get 2 questions wrong and get a 750 while in another you may get 2 questions wrong and still receive a 790. That’s a pretty massive point difference for a top scorer and is a strong incentive to retake the test and maybe get a lucky answer (or hard question that just happens to be right up your alley) on one of the retakes so that you get bumped up to an 800.
For repeat test takers not trying to get a perfect score there’s also the incentive to learn to avoid harder questions and leave the answers blank since a blank doesn’t harm your score as much as a wrong answer. This is a learned skill that someone unaccustomed to it likely isn’t going to pull off as well on their first attempt at it (and if you’re that smart and naturally talented you likely don’t need the mechanic’s assistance that much).
>there’s also the incentive to learn to avoid harder questions and leave the answers blank since a blank doesn’t harm your score as much as a wrong answer.
Sort of, not really. Sure, a blank is better than a wrong answer, but the expected value of a random guess is the same as that of leaving it blank, zero, because you get 1/4 point off for a wrong answer, but will, on average, get one answer in five right.
Of course, it’s not truly random, and test-takers could be subject to a bias–most SAT verbal questions have intentional red herrings designed to catch students relying too much on key words, for example, so it’s not impossible for someone to consistently perform worse than chance. But overall, I expect savvy-ish guessing to have at least a small positive expected value on average.
Yeah, to be honest, I’m not sure why they’re moving away from the wrong-answer penalty.
It made sense: the problem is that in a regular test, you have a strong incentive to bubble in all the answers even if you don’t get time to even look at all of them (and they are designed to do this). The penalty makes it, on average, just the same if you don’t randomly bubble in as if you do.
@TheAltar (did you see that pun coming when you picked your name?):
This is an important point, and I have the interesting suspicion that it affects people very unequally. Presumably variance has a bigger impact at the high end of the test than the middle (both from statistics, and because 750 vs 790 looks like a bigger change than 550 vs 590). And, presumably, this creates larger risks for certain populations.
If you’re an Asian student applying to Princeton based on your academic strength, those 40 points at the high end could make a massive difference. If you’re a non-Asian student applying to [Midtier College Here] on your athletics and charity work, you’re going to feel it a lot less.
Some of that is an obvious “insight” (students counting on academics care more about the SAT, duh), but it is interesting that scoring variation should matter more to groups on the edges of the exam.
@anodognosic:
The “equal value” claim is also ruined by the fact that SAT consequences are thresholded, rather than just a skill assessment. (As always, if you incentivize something it ceases to be an accurate metric.)
Students who’ve done practice tests can game the system by knowing what scores their desired colleges expect: low-average students should increase variance, high-average students should decrease it. Further, all students have some incentive to increase variance the first time around: if you score badly you can always retake, but if you score well you can submit that alone and look great.
Also, it takes specific knowledge of the test scoring to know when to answer if you’ve been able to eliminate some of the answers to a question, but you don’t have it down to one answer you’re sure of.
Here’s another of those vaguely remembered things seen on line– that oddly enough, the people who did worst on the math section of the SATs are those who were never taught the material in high school.
On the quarter-point penalty, it always made sense to me, not only because it evened out the expected value of guessing and skipping (instead of infinitely privileging guessing), but because I think it encouraged even a minimal amount of trying. Even if you are able to eliminate one wrong answer, your expected value is now positive.
Now of course, eliminating choices would still benefit you if guessing had no penalty, but the penalty reduces the effect of random luck and, at least I hope, gives test-takers a clear choice of which action to take based on their level of confidence (which should reflect their reasoning ability). This might be the sort of thing anodognosic is referring to when they say “savvy-ish guessing,” and I do think it means something (at least being able to filter out obviously wrong answers, or in math questions, working through some of the steps, if not all, to eliminate now impossible answers, etc.).
Now students will be incentivized, with five minutes left, to fill in A/bubble in an obscene picture/etc. for all their unanswered questions, introducing noise that will reduce the difference between test-takers at different ability levels.
@ Zaxlebaxes:
Maybe I’m missing something, but I don’t see how the wrong-answer penalty encourages trying.
You have, on average, the same incentive either way to eliminate wrong answers and make better guesses.
The only thing it does is reduce noise and luck. And increases the range of scores by eliminating the ~20% floor.
So why does the literature show gains from tutoring/test prep well below 83 points? Maybe the true amount is 83 points but only for those dedicated enough to work their asses off for a whole year+ taking the test over and over again? Or maybe the people taking the test 5 times have some selection bias, maybe they are lower scorers with more opportunity to improve or something?
To me, the obvious guess would be that people with such a high number of retakes don’t offer a random sample of initial scores. If you get a 600 and expected a 700, you retake. If you expected a 500, you don’t.
Even across many retakes, you presumably quit if you get a high score you’re happy with. Some people will be misjudging themselves and won’t ever improve, but the people who really did screw up the exam last time will have a strong incentive to retake.
Nor is that the only sample bias we might be seeing. Perhaps weak test-takers get stressed out and bomb, then slowly improve as they become more comfortable over retests. Perhaps students who get tutoring are particularly unlikely to improve, because they come from rich, academics-promoting families and have already gotten most of the studying/test skills gains available to them before tutoring (anyone know how free SAT prep for troubled students compares to paid SAT prep for rich ones?)
In any event, my best money is on massive sample bias between those two groups.
I’ve never really understood why the SAT wouldn’t respond to studying. I tutor the SAT, and I see people improve all the time. I’ve even improved my own already high scores over the time I’ve been tutoring.
The reading sections can be hard to improve, but the math and grammar sections both depend pretty heavily on having specific knowledge that can be taught. I have an ACT (not exactly the same as the SAT, but very close) student right now who improved his score noticeably just by learning the rules about commas and semicolons.
Because the SAT is (supposed to be) an aptitude test, and hence acts on the assumption that everyone taking it has been taught the appropriate skills. Given that everyone has been taught the necessary material, scores on the SAT should relate directly to aptitude/IQ/intelligence. When that assumption is not valid, you can have very strong gains from learning some fundamental skills. (E.g., in extremum, non-English speakers will do very badly on the SAT regardless of how intelligent they are.) But if the basic skills have already been taught, then there’s very little more to do other than teaching how the test itself works, which doesn’t affect scores much.
I feel like the ACT is a more straightforward “knowledge test” than the SAT, which is full of trick questions and things that lead you off down the wrong path.
Particularly in their respective math sections. The ACT just gives you a list of straightforward math problems and asks you to solve them. Many of the SAT ones, in contrast, are deliberately misleading.
Math was my weakest part on both, but I did significantly better on ACT Math. As I recall, 32/36 on the ACT Math and mid-to-upper 600s on the SAT Math. I applied to college with my ACT scores, along with some SAT IIs (I think US History and…Biology? Looked it up: US History, Molecular Biology, and English Literature) and APs.
Caethan,
I agree, but very few people taking the test actually have all the appropriate skills. If you had 100% of the necessary knowledge for the grammar section, you should be able to get a virtually perfect score. If you read a lot, you’ll have better intuition about which answers sound correct, but ultimately it’s about knowing the rules, either explicitly or intuitively.
The math section is somewhat more aptitude dependent, but most people still don’t know all the necessary skills. Even beyond that, it’s helpful to learn how to solve specific problem types. I’m good at those kind of tests partly because I’m smart, but also because I’ve seen most of these problem types before, and I already have a decent idea how to solve them.
However, looking at Table 6, the time gap between tests doesn’t matter, and people retaking only once show much smaller gains.
People who show a very small gain after retaking once don’t take it a third time?
I had some involvement with the redesign of the SAT a few years back and as I recall, the rule of thumb when “scaling the SAT and PSAT was that each year of the test takers age (maturation+education) within the age range of high school was expected to result in a 100 point gain in their score. Thus the SAT and PSAT were scaled to each other such that a student taking them simultaneously would be expected to get ~200 points higher score on the PSAT, since that test is designed to be taken at the end of 9th grade or beginning of 10th grade while the SAT is expected to be administered in late 11th or early 12th grade. A student taking both the PSAT and SAT at the normal ages is thus expected to get similar scores. A student taking the SAT twice a year apart would be expected to gain 100 points between scores.
This doesn’t mean that the variance is necessarily due to education or study and isn’t inconsistent with the SAT basically being an IQ test. As I understand it, clinical IQ test results for children and youths are normed according to the test takers’ age during scoring, while the SAT is not.
“A student taking the SAT twice a year apart would be expected to gain 100 points between scores.”
Hard to do if the score the first time was over 700 (I assume your gain is on each test, not on the sum).
Do you know if the norming of the PSAT was done the same way fifty+ years ago? I’m pretty sure my score went up from the PSAT to the SAT.
Makes sense, anyone smart enough to understand the broad meta-textual of the SAT, are likely smart enough to understand a few reading comprehension questions.
I just took the SAT recently (I’m a senior in high school now). I took it a year ago in December without studying at all and again last May after doing 3 practice tests the week before. The first time I got a 2170, the second time a 2250. I improved by 20 points in both Math and Reading and 40 points in Writing for an increase of 80 points overall. I also did far lest test prep than most other people at my school who retook the test. But this of course is in the 2400 scale, not the 1600 one, so the scoring may not be very comparable.
I was an example of factor#1 – tired/stressed versus well-rested:
The morning of the SAT I was sick with a cold so I took an antihistamine which made me groggy such that I was literally falling asleep at my seat during the test. So when I retook the test the score went up by over a hundred points (“Math” went from 720 to 800; “Verbal” improved by a more modest amount).
I went the other way. Time #1 I was out till 3 am and up before 7 to take the test. I took naps in between the sections, I scored 20 pts higher than my retake.
Speaking of which, this might be a good place to ask: I could swear I once read about a new technique for getting a measure of infections one has experienced over a lifetime, so you could see person A had had 10 serious infections and person B had had 20, and that this was applied to identical twins, showing that they often had very discrepant viral loads and this was probably a decent part of non-sharedenvironment. But I couldn’t remember any of the details and when I went looking for it a few months ago, I couldn’t find it after hours of searching, and it’s been bugging ever since. Does this ring a bell for anyone?
Not sure how new it is, but you can get a decent record of your history of infections by looking at the frequencies of the major types of antibodies in the blood (IgA, IgG, IgM, IgD, and IgE). For instance, a concentration of IgM antibodies for rubella indicates a new infection, but IgG antibodies without the IgM antibodies suggests an infection in the past.
I’m way out of my depth with this biology, so you should definitely double-check for yourself.
I think this is what you are looking for.
Looks like it, thanks. I don’t see any mention of twins, though, in it or in any of the citing papers in Google Scholar. Maybe someone merely suggested that it could be used on twins and I misremembered that.
The other side is that we know that some sorts of things vary culturally, eg which language you speak, and virtually everyone in a culture picks up on them.
There should maybe be a category called “obviously environmental effects.”
So for example how well a child does math is almost entirely a function of inherent ability. The teacher makes little difference, the school little difference, mom and dad teaching you multiplication tables doesn’t matter.
So when one kid ends up much better at math than the other conclude the first kid is just better at math. However, if one kid grew up attending school and learning math, while the other worked in a coal mine, and the first ends up more numerate, you know nothing about their relative mathematical ability. Obviously.
Something that worries me is the question of how different the “non-shared” environments are. If we’re talking about siblings (as some of the studies are), they seem likely to be growing up in environments that are extremely similar. Even if they’re studying people that merely live in the same country, that’s going to be a lot of similarities. Obviously, when studying a population that doesn’t have very much variation in environment, variation in environment shouldn’t be expected to have big effects.
Siblings by definition have shared environment – that is, nonshared environment is calculated as things that differ even among siblings.
Your country point is a good one, but we often talk about eg what the US government should do to improve people’s outcomes, in which case the variation attributable to different things within a country is what we should be concerned about.
There is some evidence that environment matters more in samples including very poor people, who don’t make it into the usual studies.
Doesn’t this tell us that twin studies aren’t likely to be very good at telling us how big an effect very large differences in non-shared environment have on people?
I mean, the number of twins raised, one in poverty and the other in wealth, by the same biological parents, must be zero or very near to it.
So what exactly can twin studies tell us about how poverty affects development independent from genetics?
There are some huge identical twin adoption studies that form the basis for the study of non-shared environment.
But how much variation is there among families who adopt children? I would expect, among other things, that they would be richer than the average population.
I’ve also read that adoption agencies actually put a lot of effort into matching up adoptees with “compatible” adoptive families. If that is true, then it would seriously undermine the idea of adoption as a natural randomized experiment.
One French adoption study (not involving twins) tried to find examples of 20 trans-class adoptions but could only find 18. Their data, limited as it was, suggested that IQ at age 14 was 59% nature, 41% nurture.
I think this article would interest you: http://www.nytimes.com/2015/07/12/magazine/the-mixed-up-brothers-of-bogota.html?_r=0
Basically two sets of twins were born around the same time in a hospital in Columbia. One baby from each set was accidentally switched in the hospital. One of the now mixed set of twins grew up in not a wealthy home, but in what is probably middle or working class in Bogota. The other set grew up in a tremendously poor rural farming community, their house was basically a shack, and the closest road you could travel by car on was a 5 hour walk away. So this is basically a wet dream of a twin study. The article is pretty long, but very extensively detailed and interesting.
Much of the research for the study of these 2 pairs of twins switched in the maternity ward was done by Nancy Segal of Cal State Fullerton, who is pretty much the Twin Queen of the social sciences. If you are into twins, just look for Nancy’s name and you’ll know you’re getting good stuff.
That’s not the purpose of twin studies, the purpose of twin studies is to estimate how genetics affects outcomes independent from any other non-genetic factor (excluding non-shared novel mutations).
If you measure the income of pair of identical and fraternal twins, and find that the income of pairs of identical twins is more correlated than the income of pairs of fraternal twins, then this is evidence that there is a genetic effect on income. You can then do the math and turn these correlation values into a “heritability” measure.
“Your country point is a good one, but we often talk about eg what the US government should do to improve people’s outcomes, in which case the variation attributable to different things within a country is what we should be concerned about.”
Cross-cultural variation is really bad for anyone who wants to use studies like the one discusses in the OP as evidence for natures importance over nurture*. If the split between genes and the environment is 50-50 — or even 90-10 in favour of genes — when the difference in environment is incredibly small in comparison to variety of environments that humans actually live in, then it suggests that the environment plays a considerably larger role than genes. The cultural/environmental difference between poor Americans and rich and poor Americans, African-Americans and white Americans is minuscule compared with the difference between any one of those groups and Ancient Egyptians or a tribe living in the Amazon jungle.
You suggest the cross-cultural point is irrelevant when it comes to producing different results for people within a country. This only holds true if we limit potential changes to nothing that isn’t extremely similar to the way we’re already doing things.
*I personally think the distinction between nature and nurture, or biology and culture, is a false one but I’ll stick with the terms here for simplicity’s sake.
“*I personally think the distinction between nature and nurture, or biology and culture, is a false one”
Are you a lamarckian? You believe environment can change your genes?
Well, the environment has some impact on mutation rate. If you happen to live in a very hot area, or very radioactive area, you’re going to see more random mutation.
While the environment can’t currently change your genes, it certainly can affect which genes new people end up with. If genetic engineering in humans becomes practical at a large scale, the environment could become just as important as inheritance for determining the genes of newborns. I’d say the distinction is definitely at least blurred at that point.
Uh, I don’t really care a whole lot about whether the differences between modern Americans and Ancient Egyptians are caused by nature vs. nurture. What exactly would be the impact of that on any modern policy? Whereas knowing that the differences between poor and rich Americans is broadly genetic or broadly environmental has a huge impact on what sort of policies we should choose right now in modern America.
Not everything in siblings’ environment is shared, though. Between when I reached each developmental point and when my younger sister did, my parents had a lot of time to refine their parenting methods and – it turned out – be significantly more lenient with her than with me. When we moved to a new neighborhood, I was age N and had a lot of studying to do inside; she was age (N-K) and just young enough to still roam around yards with all the neighborhood kids. Being older or younger at a given time matters.
Plus, look at all the studies about the significant effects of birth order within a family. Having an older sibling to look up to, or a younger sibling to help, matters a lot in your psychological development.
According to Judith Harris in _The Nurture Assumption_, the birth order stuff is basically bogus. If there are enough different patterns you are looking for, in any one study you can probably find one by chance. Different studies find different patterns. Pool all the studies and the patterns vanish.
I have no expertise in the subject but found her book pretty convincing on that and other topics.
Thanks, David. I saw this comment earlier and was thinking, “Didn’t I see somewhere that birth order arguments are ‘refuted’?”
I think there are a lot of weak arguments for birth order effects that try to draw out slight differences in personalities between, say, the second middle and third middle child. But the strongest effect that I continue to observe as I get older is that if there is a single “wild child” in the mix, it is almost never the eldest. Of course, the “wild child” doesn’t always stay wild, which is consistent with the idea that parenting/environment makes more of a difference in childhood than adulthood.
Also, isn’t homosexuality correlated with being the second son? My father made this observation, very unscientifically, a long time ago. He attributed it to some sort of admiring of the older brother, but now I think the theory is that it has to do with testosterone exposure in the womb.
So it’s not impossible that there are multiple hormonal effects related to birth order that are consistent with an otherwise hereditarian position.
Wency, I’m going from memory here, but while there seems to little evidence for birth order having an effect. Less Wrong has a weirdly high proportion of first borns. This proves that the universe has a sense of humor.
As I understand it, homosexuality doesn’t just correlate with being a second son, but with the number of older brothers.
Have they ruled out, among the lesswrong data, that it could be caused by the greater genetic defect rate among later-born (since western people already tend to be born late, when their parents aren’t in their genetic prime)?
How many were only children (and therefore, technically, first-born)?
Anonymous, that sounds plausible. It might be after you adjust for age of parents at birth that birth order effects disappear. Still, if there’s a homosexuality effect in birth order, can that really be the only effect?
I’m thinking of the people I know who came from good homes but dropped out of school, got addicted to hard drugs, got pregnant at a young age, etc.
I can only think of one who is the first-born — a guy who had a nervous breakdown in college, got into drugs, dropped out, and I’m not sure if he ever went back and finished.
Strangely, he is part of a pair of identical twins where the other twin avoided hard drugs, finished school in four years, got married, and got a good job. It would be a good case study for non-shared effects, and I have no idea what to attribute the difference to.
One of the LW surveys also asked for the parents’ ages when the respondent was born and IIRC the answers weren’t unusually low in average.
I am a layman here, but … siblings must have a somewhat different environment when it comes to … their siblings. There may be a divergence when each sibling is looking for their niche in the family – Peter may decide that since Paul is the mom’s boy, he might rather be the more independent/adventurous one etc. So at the end each sibling is growing up in a slightly different family. I believe there are also different traits in siblings based on order of birth etc.
Right. I know a pair of identical twins who are obsessed with any and all of their differences. One informed me his eyesight was 20-22 and his brother’s was 20-24. One wants to be an engineer the other an actor.
So, sibling rivalry sometimes makes twins reared together more different in certain ways than if they were raised apart.
For example, Horace and Harvey Grant are identical twin 6-9 basketball players. In high school Harvey was the shooting forward, so Horace was, despite his slender frame, the power forward. If they had been raised separately, they likely would have both been shooting forwards at different high schools. (This actually worked out to Horace’s advantage since he enjoyed a longer NBA career, perhaps because he had to stretch himself to learn a different role in high school, whereas most future NBA star 6-9 guys are shooters in high school.
Judith Harris has a later book, _No Two Alike_, which is about the sources of difference in identical twins reared together, including one case of siamese twins who were quite literally together to adulthood–and had noticeably different personalities.
There is some evidence that environment matters more in samples including very poor people, who don’t make it into the usual studies.
That doesn’t sound great, if the results of such studies are being used to create government policies and programmes for school-going children of the poor/parents living in poverty/poor high school drop-outs etc.
Variance component estimates are relative to particular populations and environments, yes. So better environments will tend to yield smaller environmental effects.
Does this matter? I guess it depends on what you want to apply it to. If you’re interested in political or sociological matters, or in improving the general weal, this isn’t really a bad thing: it tells you that you’re going to have a hard time making things better with the simple obvious interventions. eg if schizophrenia turns out to have a small shared-environment effect, you’re probably not going to reduce rates meaningfully by some lead remediation or anti-fridge-mother educational programs, you’re probably going to have to adopt much more radical approaches.
One notes that all twins — identical or fraternal — shared an environment for nine months at a time where the influence was probably large.
That’s a point I was going to bring up, with the additional complication that the uterine environment is also partly genetic.
But even in the womb, one twin typically gets a better position.
I had a question about this shared environment stuff. What happens if you measure the heredity of “what language do you speak at home” or “do you use chopsticks when eating most meals”? Intuitively, you should get near 100% shared environment effects, right? Or am I misunderstanding? (Surely chopstick use is not genetic, right?)
Did anyone try to do this, just as a sanity check of the measurement methods?
You would get the right answer. Consider https://en.wikipedia.org/wiki/Falconer's_formula 2*(r_mz – r_dz). If you look at the correlation of monozygotic identical twins using chopsticks, it’ll be r=~1, since everyone in a household uses chopsticks or none; similarly, dizygotic fraternal twins will also correlate ~1 (or at least, I assume as much), so then it’s h^2=2*(1 – 1)=2*0=0; if h^2=0, then shared+non-sharedenvironment=1. You could also look at adoption studies: if the adopters use chopsticks, so will the adoptee, and likewise for nonadoptees, so h_n^2 = r_sib – r_ad = 1-1 = 0. So the formulas would get sensible answers in the case of parentally-determined stuff.
Thanks. Followup question: if cultural things like chopsticks are shared environment, what about the Southern US “culture of honor” I keep hearing about? Would some notion of “honor” have a large shared environment component? Or would you say that there’s no such cultural effects at all (a pretty remarkable claim, if you ask me)?
What about effects like “were you ever a member of a gang”? Surely that depends on whether there were gangs in your neighborhood, yes? So can we predict some strong shared environment component?
Basically what I’m asking is, do we EVER see a large shared environment component in ANYTHING? Because if not, something seems wrong.
E.g. in behavioral genetics studies of politics, partisan affiliation (Democrat vs Republican) shows much bigger shared environment effects than ideology measures.
Similarly accent and religious affiliation show high shared environment, while vocabulary and intensity of religious practice show higher heritability and lower shared environment.
See this chart breaking down political traits by genetic, shared environment, and non-shared contributions:
http://www.cell.com/action/showImagesData?pii=S0168-9525%2812%2900111-4
There’s a famous pair of identical twins raised apart known as the Nazi and the Jew twins due to their different political environments when they were children:
http://www.unz.com/isteve/identical-twins-separated-near-birth-the-jew-and-the-nazi/
They were, however, extremely similar in many ways.
“What happens if you measure the heredity of “what language do you speak at home” or “do you use chopsticks when eating most meals”? Intuitively, you should get near 100% shared environment effects, right?”
Yes, except insofar as siblings differ on those.
Okay, but are there actual studies that show shared environment effects in these things (or in *anything*)? If not, why don’t people conduct such studies? Chopsticks are boring, but what about other cultural effects like honor, politeness, respect for authority, etc.? (Surely those are different in Japan than in New York, and surely that difference is not 100% genetic, right?)
Jacob linked to a mega-meta-analysis of all pre-2012 twin studies below. There are shared environment effects found for many, many traits.
The OP exaggerates here:
“studies of many different traits tend to find that 50% of the variation is heritable and 50% is due to non-shared environment”
Non-shared environment tends to be larger than shared environment, but the latter is not uniformly zero.
“Surely those are different in Japan than in New York”
Yes.
Most all behavioral genetic studies are within a country, but when you include international datasets heritability goes down (because of the additional of international environmental differences). You can see that in publications of the Social Science Genetics Association Consortium (which use samples from many countries in Europe and North America).
If you do a behavioral genetics study with data from different times/cohorts, you will also get bigger environmental effects and lower heritability, for the same reasons.
Yes, this is why I thought it was a mistake when Scott took the studies he was looking at as evidence that it was pointless to try to change people by reforming society; surely one has to look at different societies to see how much difference changes in societies can make. And most of the studies don’t do that.
Carl writes:
“If you do a behavioral genetics study with data from different times/cohorts, you will also get bigger environmental effects and lower heritability, for the same reasons.”
That’s related to the Flynn Effect, where it turns out that IQ test scores tend to go up over the decades at a fairly steady pace.
Twin studies, especially identical twins reared apart, work very hard to look at nature and nurture over Space, but it’s slower to look at things over Time, which the Flynn Effect suggests is surprisingly influential.
I’m too tired to say anything intelligent, but I’d be remiss for not posting the Meta-Analysis of Twin Correlations and Heritability (http://match.ctglab.nl/). It’s a meta-analysis of all (and I do mean all) twin studies published up to 2012: “Currently the database includes information from 2748 papers, published between 1958 and 2012, reporting on 17804 traits on a total of 14,558,903 twin pairs. Have Fun!”
How many of the 14,558,903 twins pairs are overlaps? What would be the record for most number of studies a twin pair participated in? I wish I knew.
It may be a controversial addition to your list, but free will is another thing that would show up as “non-shared environment”. I don’t especially want to get into a free will debate, but assuming it’s real, Im guessing it wouldn’t end up on the nature side.
That’s covered under “biological random noise”.
Oh, yes. It is, however, very unpopular among people who do studies. I have actually read one saying that he doesn’t believe in free will because you can’t have a science of something with freedom. . . .
Little gap of logic there.
I think that viewpoint is misguided, but there’s a kernel of truth there: if free will is the ability for people to be an uncaused cause of some of their actions, then to whatever extent a person acts on the basis of free will, those actions can’t be explained on the basis of prior causes.
For instance, suppose 100% of the difference between successful people and unsuccessful people is that the former are taught the value of hard work. Well, that’s not free will. That’s a prior cause: the former were taught something the latter were not.
As I quoted in a previous thread concerning the medieval thinker Peter John Olivi:
The bottom line is that to the extent people’s choices are determined by free will, they are unpredictable and there is no outside intervention that can be imposed to improve their outcomes.
I think we’d need to taboo “free will”. People use it in radically different ways.
Some people (particularly from the humanities) treat it as a magical thing, free from the influences of physics/quantum physics and cause/effect where the magical soul impinges on physical reality and because of their version of “free will” the universe cannot be deterministic.
Others just mean that your actions are not practically speaking utterly predictable based on easy to compute simple rules.
If you’ve spent too long trapped in a room with true believers in the former then you can become quite dismissive of every mention of the term.
Measuring sources of variation shouldn’t put a ceiling on what interventions *could* do. If you start with a population in which everyone’s being poisoned by lead in equal amounts, lead explains very little of the variation of intelligence. If you clean up all the lead for half the population, lead exposure will explain a fair bit of variation in intelligence. If you clean up all the lead for everyone, lead exposure will once again explain no variation.
On the other hand, it puts a floor on the level of intervention needed to achieve noticeable change. If a variable affects some outcome but only when it’s moved beyond the range of normally encountered values, then you have to perform some radical changes to make any use of that information.
Yeah, lowering lead exposure isn’t really very radical but it’s not an example of a social intervention, the sort that nurture people really want to believe in.
Not entirely sure who you’re referring to as “nurture people”, but I suspect it shares a lot of overlap with people who are really concerned about:
a) poverty
b) pollution
In addition, lead is a non-racist explanation for racial disparities in intelligence and social well-being, and it can be blamed on evil corporations. It seems to me that lead remediation is exactly the sort of social intervention “nurture people really want to believe in”.
There’s quite a few liberals/leftists who aren’t comfortable with theories that highlight both a demographically measurable difference in IQ and significant downstream effects of that difference even if it’s a difference that can in fact be blamed on racism and the externalities of the private sector.
I’m not one of them, though.
Don’t forget gut flora!
That would have been my contender for ‘3e’. If you consider the human being as a ‘hotel’ for nine times as many ‘cells’ as those human body cells then the total (human and gut flora) genetic profile could easily vary between twins. Diet can lead to a different gut microbe profile, as can environmental factors like infections and poor hygiene, who you live with, or even (especially) courses of antibiotics.
The part that I bolded is the part that I take the biggest issue with. Correct me if I’m wrong, but don’t most of these studies sample people not just from the same country, but generally from within specific geographical regions, and often from within a relatively narrow range of socio-economic status and similar factors?
If so, then couldn’t the low values for shared environment simply indicate that the samples didn’t capture much shared environment variance? Wouldn’t this make these studies practically useless for determining whether or not “fixing society” could improve people’s life outcomes? These studies could effectively be controlling for societal factors.
A bit of anecdotal data on cultural differences: I recently moved from a city in the American South to a city in the American Southwest, and I have encountered a lot of noticeable differences in the way that people act. In particular, I only recently “got” what people mean when they talk about “Southern friendliness,” now that I’ve spent a lot of time in a culture without that trait. I have a very hard time imagining that these differences are due to genetics. Despite speculations about Scotts-Irish ancestry and the like, the ancestry of various regions is very heterogenous when you examine the data. For example, the population of the city that I grew up in, New Orleans, is about 60% black, 30% the same mix of European immigrant descendants that you find in every Atlantic port city (resulting in an accent that has been described as “Brooklyn on Valium”), and 10% Latino or Asian.
Shared environment can vary even within a homogeneous group for factors like whether the mother stays home with the kids, how strict the parents are, what extracurricular activities the kids do. Parents tend to worry a lot about making the ‘right’ choice on issues like this, and the question is how much effect it even has.
In theory, but how much do they vary in practice within a homogenous group? Even comparing different generations in the same culture you find differences in parenting practices that utterly dwarf the differences between modern parenting fads. To slightly paraphrase Jeff Foxworthy, “Now you are supposed to give your kids a ‘time out.’ Our parents took time out of their busy day to whup our ass.” On the other hand, it is common to hear testimonials from people who grew up as late as the 1980s saying that their parents allowed them to ride their bikes across town when they were in single digit ages.
It isn’t just parenting either. All sorts of things change between cultures or just over time within the same culture. To name one example, I gather that physical bullying was a lot more common a few decades ago even than when I was in elementary school in the 1990s. For a comparison between contemporary cultures, in the past few years most of my Facebook friends living in rural Louisiana got married or even had children in their early-to-mid-twenties (the youngest of them got married at the age of 23). Nothing that they posted gave any indication that this was considered unusual where they came from. That’s an almost certainly social environmental effect with a very large impact on life outcomes, but would these studies have captured it?
That may have been true of earlier studies, but there are now a number of big twin samples which are nationally representative. Several European countries have twin registries which recruit basically all twins born in the country. In the US, many twin samples can be regarded as representative of the populations of particular states at least and there are a couple of national samples as well.
Adoption studies have a restriction of range problem in that the worst kind of parents are usually ruled out by adoption agencies. Parents of adoptive children tend to be like Steve Jobs’ adoptive parents: stable solid citizens (while his biological parents were brilliant but unreliable). Adoptive parents aren’t necessarily wealthy or smart, but they are seldom impoverished or vicious.
I knew a reasonably rich family that adopted a child and then gave her back when it turned out she was merely bright and pleasant as opposed to brilliant and studious like their genetic child. It was weird, because they were an otherwise amicable and warm family and it was completely out of the blue.
The daughter spent a while on welfare raising her child, who she absolutely fawns over. So, she didn’t pick up much from her adoptive parents.
I would expect that adoptive families are unrepresentative of the general population in a number of other ways. Most obviously, they almost certainly have a higher rate of infertility, which correlates with age at first marriage and who knows what else.
Does anyone know how common scenarios like the one God Damn John Jay describes of an adoptee being returned by the adoptive family for whatever reason are? Because if it is at all common, it seems like it could pose serious problems for adoption studies.
On a semi-related note, I found an adoption agency’s “Guide to Connecting Adoptive Families with Waiting Children” through Google, and just from skimming it it looks like the process that determines which kids end up with which parents involves a lot of non-random factors.
There was a famous story about a family that gave up a severely mentally ill child they adopted from (Russia?). I think I have heard similar stories out of Popehat as well.
What’s the dependent variable that you are interested in? Clearly, environmental (social) factors are more relevant in explaining income than in explaining criminal behavior, and criminal behavior in turn is probably explained to a larger extent by environmental factors than talents, personality or intelligence.
But maybe I’m missing the point?
“Clearly, environmental (social) factors are more relevant in explaining income than in explaining criminal behavior”
Heritability of income in men is 0.58, heritability of crime is more like 0.5
the mechanisms behind the heritability are social, very likely in the case of income, somewhat less likely in the case of crime.
When you see “heritability” in a scientific or medical paper or usually in these comments, it’s talking about genes or (less likely) prenatal environment or weird epigenetic stuff, not social mechanisms — those are split between shared and nonshared environments. It’s measured with twin studies or other protocols that try to control for everything that happens after you’re born.
This means among other things that the more the thing you’re looking at depends on stuff your parents taught you or money you inherited from them, the less heritable it is. It’s counterintuitive, yes.
Thanks, Nornagest. I did not know that this is the common terminology.
Pre-natal environment is not included in heritability. This is not negotiable. There is a correct answer and an incorrect answer. Adoption studies (but not twin studies) will include it in their measurement of heritability, but that is systematic measurement error, not a variant definition.
When it comes to inherited epigenetic effects, the definitions break down and there isn’t a clear correct answer.
“This means among other things that the more the thing you’re looking at depends on stuff your parents taught you or money you inherited from them, the less heritable it is.”
Not necessarily – take math ability; your parents may “teach you” math (stuff your parents taught you) but maybe they did so because they themselves were (genetically) predisposed to be good at or interested in it. In this scenario is this more or less heritable? We don’t know unless we randomly assign some parents good at math the task to not teach their kids math and then look at how their kids compare to their counterparts whose parents had no such restriction.
The trick with humans is that we have an “external phenotype” – i.e. our genetic predispositions compel us to “create” a particular environment that we subsequently measure as environment and contrast it with “genetic heritability.”
the swedish twin study on income to which scott links is very interesting but limited (1) to the citizens of the very egalitarian swedish welfare state (2) to the extent that very rich people were excluded as outliers, although the questions which factors make people super-rich seems quite relevant. I’m sure the results would be different for the United States, not to mention the entire world, including countries such as Switzerland (58,149 int$/cpt) and Congo (729 int$/cpt). Maybe this seems trivial, but whether income heritability is 0.58 or ridiculously small depends on what context is implicitly controlled for.
Is this true? Eg 60 years ago, being gay was illegal, now it isn’t. So someone could be criminal under one set of laws but not under another even while still doing the same thing.
Was actually being homosexual illegal? Or just performing homosexual acts, regardless of one’s orientation?
There’s variation in how much people prefer homosexual acts, so that affects the odds of being convicted of homosexual acts. Or were you just looking for more specific language?
The laws on which drugs are illegal vary according to time and place, so that’s another example of the same acts being criminal or not depending on circumstances.
Criticizing one’s own government is also variably criminal, and it wouldn’t surprise me if there’s a genetic component to how likely people are to criticize their governments.
I’m not taking Tracy’s comment at face value (or rejecting it out-of-hand) because I’ve heard before some of the more curious stances on homosexuality that come from Protestant Americans. Compare the Catholic “you’re okay if you don’t act on these impulses” vs opinion of at least some American Protestants/Evangelicals “if you feel these impulses you are damned to Hell, regardless of whether you act on them or not”. So it wouldn’t be too far from the realm of possibility that there would be a law forbidding someone to be internally homosexual, regardless of sodomy.
«Catholic “you’re okay if you don’t act on these impulses”»
I was raised Catholic in Portugal. And we had to say things like “I have committed sins either by action, inaction or thought”. Jesus himself in the Bible says to take off your own eyes if you look at a woman the wrong way, because it is better to go to heaven without some part of your body than not going to heaven at all. So they all punish thought crimes. I don’t think religions are as easy to stereotypes as you do there.
@Ricardo
Per the CCC:
This is not a condemnation of those who suffer homosexual urges to Hell for having them. They’re in the same boat as those who have other kinds of lustful thoughts.
@ Anonymous:
And that boat in sailing to hell, unless they repent and ask for forgiveness in the name of God.
Which, of course, gay people are not going to be inclined to do insofar as they don’t think they’re doing anything wrong.
Consider the possibility of the existence of people who do not find their natural inclinations to be morally right.
@ Anonymous:
I didn’t say anything to rule out that possibility.
I said “insofar as” they they don’t think they’re doing anything wrong.
Of course, in order to be convinced that their natural inclinations are wrong, society would have to send that message to them. Which seems incompatible with a message of warm acceptance.
“We’ll accept you and consider you morally good so long as you deny your fundamental sexual impulses.” It should be obvious why that message has little appeal.
That’s not to say that, in and of itself, makes the Catholic position wrong. Maybe the truth is not very pleasant. Many non-religious people think a similar thing about pedophiles: that they don’t have any control over their “orientation”, but that they are “called to chastity” and just have to repress it because acting on it would be immoral. Nevertheless, they’re naturally going to fear and mistrust pedophiles because they recognize that they are cursed with a greater tendency to do evil.
And that boat in sailing to hell, unless they repent and ask for forgiveness in the name of God.
Not necessarily. The thoughts must have the assent of the will. Trying to control your thoughts soon reveals that thoughts frequently occur without its consent. The term for this is concupiscence.
And yes, there are Protestants who think that concupiscence is sinful. Not orthodox Catholic teaching though.
@Vox Imperatoris
OK, that makes sense.
Idly, what’s your standard of classifying pedophile behaviour as immoral, but homosexual behaviour as not immoral?
@ Anonymous:
One of the two is both immoral and illegal because it’s psychologically harmful and non-consensual.
The other is neither one of the two. Maybe you can come up with some argument that it’s harmful and therefore immoral (I disagree), but it’s definitely consensual.
How much do we learn from studying doctrine about how a religion works out in practice?
Oh dear, is that question a variation on genes vs. environment?
“One of the two is both immoral and illegal because it’s psychologically harmful and non-consensual.”
Our legal rules may treat it as non-consensual, but in real terms that depends on the particular case. Psychological harm as well.
Mencken mentions somewhere that he lost his virginity at age fourteen with a girl of the same age. Perfectly normal behavior in many societies. As of 1880, age of consent in the various U.S states was mostly either twelve or ten. Twelve for girls and thirteen for boys, plus some evidence of puberty, in traditional Jewish law (maybe twelve and a half and thirteen and a half–I’m going on memory).
And that was the age of adulthood, at which they could marry without parental consent. A girl could be married off younger than that by her father, but had the option of withdrawing from the marriage when she became an adult.
@ David Friedman:
I’m not saying that our age-of-consent laws are necessarily perfect or that you’re an evil rapist for sleeping with a seventeen-year-old.
We can debate over where the line ought to be drawn. I think eighteen (which is not the line in the majority of US states) is fairly ludicrous.
But strictly speaking, pedophilia refers to attraction / sex with prepubescent children.
In any case, if you want to argue that twelve-year-olds can meaningfully consent, and the sex is not harmful to them, then if that’s true there’s nothing wrong with it. I don’t know if it is true, but that’s the criteria I have.
not a lawyer, but I there’s no way you’re wrong.
but still, that’s hairsplitting
I thought it was doing things that was illegal. My apologies for the unclear language.
“I haven’t heard other people/experts in the field talk about this nearly as much as I would expect them to if it were true”
Steven Pinker makes many of these points in The Blank Slate.
Yeah, my impression of the situation from reading Pinker’s book is basically exactly the situation that Scott expresses in this post.
The non-shared environment contributes 50% of variation, just as much as genes. That means things like your friends, your schoolteachers, and even that time you and your twin got sent away to separate camps must be really important. More than enough there to continue worrying about how society is Ruining The Children, right?
You are summarizing an important aspect of the zeitgeist among educated people of a certain sort. There’s a tremendous belief that there must be identifiable Best Practices by which we can improve human behavior and thereby remedy all sorts of problems. But if the non-shared environment is less susceptible to control than we were led to believe then there is less reason for people to worry too much about following the recommendations in a click-bait article that begins with “According to a recent study…”
50% genes, 26% noise, 18% luck, 9% band camp, and 2% butterscotch ripple.
Consider my priors updated.
Why not more in the “genes” bucket? Genes that differ between twins are still genes.
Not if the genetic differences are caused by environmental differences.
You mean like living in the Chernobyl exclusion zone or next to a coal mine?
I don’t think it’s an error to classify factors like what kind of boss you have or how vigilant the cop in your neighborhood are as non-shared environmental. Nor is it wrong to look at a difference in income and conclude that genes do not 100% control professional success.
They are definitely non-shared environmental variables, but they have zero effect on ability yet huge effect on outcomes. This is why the questions need to be posed carefully. If you want to measure propensity to commit crime, the presence of an alert cop has no effect on somebody’s propensity to commit the crime (assuming they aren’t aware of the cop’s presence), but it has an effect on your ability to measure criminal activity. If you want to measure the rate at which people from a given set of circumstances get arrested, the presence of an alert cop is a huge part of the non-shared environment.
The bad boss might be relevant if the goal is to investigate ability to navigate office politics, but it’s irrelevant if the goal is to investigate ability to put together a technically sound product (assuming, for the sake of argument, that the political situation is fine up until the point that the work is a success, at which point the boss swoops in to steal). If you want to measure technical ability, the presence of a bad boss who steals credit is measurement error (if you’re measuring technical ability based on professional outcomes). If you want to measure interpersonal skills in an office environment, the presence or absence of a bad boss might be an independent variable in the experiment.
And if there was some outward, easily perceived marker, which policeman were trained correlated with danger, and therefore were more alert ….
It’s fair to conclude that genes do not 100% control professional success even if they did 100% control ability. The race is not always to the swift; that’s not a new finding.
I’m not sure whether finding out that it’s harder to get evidence against 100% genetic control of ability should make me update towards it being likely.
May it be the case that the shared environment is just very similar for different people or close to the top of its efficiency on the values measured? Let’s imagine a society where all your life outcomes depends on whether you do or do not know how to prove the Pythagorean theorem (example by A. Markov). If all the children are trained for all their life to prove it, the differences will be almost exclusively genetic, and if it’s an obscure fact no one cares about that is only learned in a few rare math clubs, it’s almost exclusively shared environment (the math club you went to). Most of the things we measure are by coincidence the things we care about as a society, so I expect them to be more genetic than average.
This all seems consistent with Gregory Clark’s research from The Son Also Rises.
TL;DR: In individual instances of inheritance, luck and genes are roughly in the same order of magnitude. Across a whole lineage, it converges on like ~80% genes.
“And it would make the opportunity to
changefilter those outcomes through geneticengineeringdiscrimination even more tempting and much cheaper.”Are you trying to say that reality is racist?
Reality is inherently unfair. It’s not always about race, of course, unless we make it so.
Still, there is no end of excuses for people (whose terminal values are a mystery to me) to choose racism as a heuristic.
>Reality is inherently unfair. It’s not always about race, of course, unless we make it so.
Sure. I inferred racism due to a combination of genetics + discrimination (sexism doesn’t quite fit), though I admit, Multiheaded’s post is highly compressed.
>Still, there is no end of excuses for people (whose terminal values are a mystery to me) to choose racism as a heuristic.
Can you unpack that a little? Racism is one of the easy heuristics, because of how easy it is to tell apart different races.
I think multiheaded is talking about Gattaca-type stuff, not racism per se.
I always thought Gattaca was a stupid movie, though. Just because some people with the rare heart condition don’t die, doesn’t make it a good idea for the space company to hire them to fly the ships over equally qualified people who don’t have it!
Need an epilogue to the movie where he’s flying the ship into land, has a heart attack at the wrong moment and everyone dies.
The problem with genetic discrimination is that it’s always way harder than non-genetic discrimination.
Let’s say I want high-IQ employees. I can either wait twenty years until we truly understand genetics of intelligence, make all my employees give me a saliva sample, pay for/wait for a whole genome analysis, and end up with something that, assuming all of our discoveries are correct, correlates somewhere between 0.5 and 0.8 with the employee’s IQ.
Or I can just have the employee take a quick IQ test and get something that correlates near-perfectly with the employee’s IQ. Sure, that’s legally iffy, but it’s hard to imagine a legal regime where the IQ test remains illegal but the genotyping isn’t.
Does anyone have an example of genetic discrimination that doesn’t fall victim to the same problem, other than the GATTACA example of people at risk of heart disease? (and honestly, a world where people prone to sudden cardiac death don’t become pilots doesn’t seem like much of a dystopia)
It’s only illegal if you can’t prove it works.
From Griggs v. Duke Power
Yes, the people arguing “liberal conspiracy” and complaining about how this is unreasonable need to keep in mind that these were the kinds of “objective tests” being used to deny people the vote.
Also, that Duke Power is a public utility.
Now, I think private businesses shouldn’t be liable for discrimination suits, whether it’s for “disparate impact” or “plain old KKK racism” . But then intention is clear: if only white people like the idea of soccer, they don’t want employers to set some policy like “we only hire people who like the idea of soccer” and then claim they weren’t intending to discriminate against black people—since whether they intended it is virtually impossible to prove. If it has a disparate impact, it should be related to the actual job qualifications.
I think that’s a perfectly fine rule for the government or public utilities.
The Supreme Court even recently ruled that New Haven was unconstitutional in its disparate treatment of white firefighters. The claimed basis was the city’s fear that promoting better-qualified white firefighters over black ones might cause a disparate impact (unrelated to job qualifications). But the Court held that the city had no evidence of an impermissible disparate impact and therefore couldn’t discriminate against white people for that reason.
While I’m not a lawyer, I am absolutely certain that if some insanely rich person, say a Michael Bloomberg*, bought out all the shares of Duke Energy and said “we’re going Sole Proprietorship, de-list us from the NYSE”, the Supreme Court isn’t going to say “well, in that case, Title VII doesn’t apply to you.”
Also, amazingly dick move in conflating the tests used at polls with what was used by employers in hiring and promotion determinations. Aptitude tests could be abused (Duke already had an outstanding policy prohibiting employment of blacks outside of certain areas, which should have been sufficient to trigger Title VII), but they were a valuable tool to help disadvantaged types, who would not (and even today, do not) have the resources to get degrees. Speaking from experience, it is emphatically not superior to have high-paying jobs go to people on the basis of mommy and daddy buying a degree in underfuckingwater basketfuckingweaving, while more experienced, more capable employees languish in low-paying jobs.
*I honestly don’t know if even he, or anyone, has got the money to do that.
@ nyccine:
Clearly not. And I explicitly said that I disagree with the court insofar as I don’t think that the government ought to be able to tell private corporations whom they can hire. I’m just pointing out that in the case at hand, we have some kind of typical quasi-public abomination.
I don’t disagree that aptitude tests could be very useful and beneficial to minorities and other disadvantaged people.
But if, in a particular case, a certain group of minorities can establish that a test is causing a disparate impact on them which is not justified by the actual job qualifications, it seems like the test is not benefiting them.
@nyccine
Note that Griggs struck down a (high school) degree requirement as well as an aptitude test. Both are supposed to be subject to a burden shifting framework — the plaintiff must demonstrate a disparate impact and then the burden shifts to the defendant to show a demonstrable relationship to job performance.
Why the common takeaway of the case seems to be: no aptitude tests ever, and any degree requirement, no matter how nonsensical, is perfectly okay, is a mystery to me.
Brad, you might equally well ask why so many people cite Griggs when it is a dead letter, superseded by (identical) legislation.
One possibility is that people have no idea what they are talking about. Another is that conventional wisdom is correct in identifying relevant precedents, much more accurate than reading the law; indeed, conventional wisdom often trumps the written law. A third possibility, which might explain why people emphasize the judicial decision, is the criterion of proof is the whim of the judge, and that the conventional wisdom is a good summary of what that whim will be.
Because a lot of companies either implicitly or explicitly require high school or University degrees and everyone accepts this, while if a company plopped a page full of Raven’s Progressive Matrices onto the table people would be angry about it.
Just because something is illegal doesn’t mean it will be enforced, and by and large lawyers and judges were people who liked school and don’t sympathize w/ dropouts.
“Public utility” doesn’t mean what you seem to think it means. “Public” means “trades stock on the market” instead of “only owned by one entity,” not “created by the government.” No part of Duke has ever been owned by the government, it is regulated by the government only insofar as any other corporate entity in its position would be, and it’s only dealings with the government as a corporate entity are to provide electric service to government buildings, but to claim this somehow makes it a “quasi-public abomination” makes me think you didn’t bother to do more than skim a wikipedia article.
The Court made it perfectly clear that all such tests were suspect, and would only be allowed if they could prove they were necessary, i.e. some government bureaucrat, who had absolutely no experience in your field, and likely no experience in anything other than politicking, was going to dictate to you if any given test was “necessary.” You’d have to be stupid, or suicidal, to wander into that landmine.
Why college degrees? Because neither Congress, nor the Court, has told businesses it’s racist to use them as a screening criteria.
You are confusing this for a “publicly-traded corporation” (or something).
Of course, Duke Energy (what it is called now) is a publicly-traded corporation. But it’s also, at least in the state of North Carolina, and as far as I know in the 1970s as well, a public utility. And that means that, despite being privately owned, it has a government-granted monopoly on power service, in return for “rate regulation”: having its prices set by the government.
It’s not technically a government agency. Neither is the Federal Reserve.
But it’s definitely centrally planned.
What do you think about the East India Company? In some sense, it was a private company. But it was kind of also part of the government of Britain.
There’s no need to be so hostile (but that seems to be your usual M.O., so whatever).
I’m not denying that there are “chilling effects” as a result of ruling like this. But as brad pointed out, they also ruled the high school degree requirement illegal.
If corporations got a whole lot more use of IQ tests than out of degrees, they’d probably use them despite the chilling effects. There would at least be a lobby.
I don’t agree with non-discrimination laws and think they are counterproductive. But I also don’t think they are Ruining Everything, as seems to be the narrative with some people.
The *perception* and practice in industry is that “a demonstrable relationship to successful performance” is a nearly impossible bar to meet in practice and that, effectively, the use of I.Q. tests for hiring is illegal. It seems obvious that IQ tests have a demonstrable relationship to IQ which has a demonstrable relationship to the ability to do a large variety of jobs, but I guess courts prefer the argument that “the job does not involve this specific task found on the test”.
It’s certainly possible that this perception is wrong – I don’t know if there is any actual case law about using I.Q. tests to hire, say, programmers. But if so, there’s a trillion dollar bill lying on the sidewalk, because they would probably work a lot better than most of the unscientific subjective bullshit that we use to evaluate job candidates today (most of which, at least in my field, boils down to really poor, uncalibrated IQ testing anyway).
And besides the economic cost, the less real information employers have about employees, the more they rely on proxies like college reputation, connections, “cultural fit” or outright bigotry, all of which are “more unfair”.
It is, however, common to use actual programming tests (as in, you give the candidate a problem and they have to write some code to solve it) when hiring programmers. I’ve taken a number of them myself during job interviews. Presumably that isn’t banned by Griggs because there is an obvious relation to the tasks required by the job.
@ actinide meta:
I’m not saying this is necessarily wrong, but does someone have a good source on this, such as systematic interviews with HR people or subsequent case law where someone has tried it and gotten burned?
Because all I’ve heard is conservative-oriented speculation. And I’m not saying people aren’t entitled to speculate; it’s a reasonable hypothesis. But is there confirmation?
However, I’ve also heard some potential downsides of IQ tests. For instance, I’ve heard someone presenting part of the Koch brothers’ concept of “market-based management” construct a graph that has two axes: ability and integrity, with integrity being something like honesty and hard work. You have four quadrants, then: high on both, low on both, high integrity / low ability, and high ability / low integrity.
Now, he didn’t bring up IQ tests in particular, but he made the argument that, of course, the ideal candidate is high on both. But the next highest is high integrity / low ability. Then low on both. High ability / low integrity is dead last.
An IQ test may show you have high natural ability. But it doesn’t prove you’re not a lazy slacker or a cheater. A college degree, to some extent, does.
I suppose they could give both. But there was also a recent case where a police department refused to hire people over a certain IQ level (I think 125). Not quite because they just want dumb brutes, but because smart people get bored and leave, wasting all the money they spend on training. So they’re definitely using IQ, but for the opposite reason.
@NN – Yes, programming tests are presumably legal. And probably they are very useful in some contexts, e.g. when the applicant stream is low quality (see FizzBuzz) or when the position demands specific and widely available skills (RoR web development). IMO they are not so good for e.g. hiring senior systems engineers to do hard and novel things – to discriminate the top people you want from merely great people, they would have to be *very* hard; they would have to be very carefully designed to not unnecessarily reject candidates who don’t happen to have some particular knowledge or skill that they would pick up in week one on the job; and different candidates will be disadvantaged to wildly different degrees by being forced to work without their usual tools (e.g. you may wind up rejecting the guy who normally uses a dvorak keyboard). In principle with enough resources I think someone could develop, and calibrate with real data, a “programming IQ test” that would work better than an IQ test, but that’s asking for a technology we don’t actually have today. In any case, IQ would be valuable additional information in all these cases, and if it was widely used it wouldn’t be expensive (you could take a test once and just put a link to the results on your resume). At the least, it would be worth trying to use it and see what the results are like!
(and, no, if IQ tests were available I would not hire software teams based on an IQ threshold and nothing else; please let’s not spend time on that strawman)
@vox – My comments about perception and practice in industry are based on my personal experience in industry as a startup founder and as a hiring manager, and from talking to other people who actually do this stuff, not from political arguments. Obviously I haven’t been everywhere and done everything, but I think I can confirm that the perception is widespread. A few seconds of googling provides this (http://www.stoel.com/labor-and-employment-law-alert-14) which seems pretty consistent with how I’ve been advised in the past – basically lawyers won’t tell you that you will lose a lawsuit with certainty, but that it will be an expensive crapshoot, and that protecting yourself with prior research will be expensive and might not work, and why don’t you just hire people from Stanford and MIT like everyone else? They all have high IQ.
It is, of course, a matter open to debate how much efficiency is being left on the table by current hiring practices, and how much IQ tests would help. My guess is that the first is astronomical and that the second is probably not to be sneezed at. Unfortunately most of the evidence I have for that assessment is either subjective, confidential or hard to talk about anonymously, and I’m not sure how the world will be much better if I convince you.
Here’s one piece of objective evidence: compensation at startups and bigcorps is generally highly backloaded through options, restricted stock grants and internal raises. Some of this is because of a bilateral monopoly situation that arises from an employee gaining “firm-specific human capital”, but a lot of it is because we don’t actually believe we know what someone is worth when they are hired.
There is also compensation compression due to poor information – firms usually don’t have to pay their best people what they are worth, because other potential employers don’t know. Most people believe there are programmers with 10x median programmer productivity; there are very few programmers making 10x the median programmer salary.
@ actinide meta:
Thanks for the source and your opinion.
I note that it dwells more extensively on the pitfalls of tests like Myers-Briggs, which is…pretty reasonable. And it also highlights that much of this is fear of being sued on the basis of disability law. I think disability law is far more economically harmful than any side-effects of laws against racial discrimination.
Mainly because laws trying to outlaw racism are for the most part laws telling people not to do what is already economically inefficient. But the ADA is a direct command to do that which is economically inefficient.
I guess I am dubious of the marginal value to employers of IQ tests, given access to college degrees. I see what people who don’t want to go to college get out of it. But sure, maybe there’s a lot of money on the table.
Certainly, there’s “astronomical” room for improvement in information here. But it’s mostly of a sort that we have literally no way to provide. As you concede.
actinide meta,
Extremely Interesting. Do you know if firms in other countries (such as Singapore) use IQ tests as part of hiring decisions when picking programmers?
@actinide meta
Your link does not advise employers not to use aptitude tests, it advises them to make sure that the aptitude tests actually do what they want them to do before they use them. That seems like fairly good advice, regardless of the current state of discrimination law.
The link does take a strong stance against personality testing, and against Meyers-Briggs profiling in particular. But that’s presented as a side effect of the fact that there’s very little evidence that it works.
Use of IQ tests is not so rare as you seem to suggest. Vox Imperatoris references Jordan v. New London (at least, I think that’s the intended reference; the case is from 2000, but the facts fit) where an applicant was turned away for scoring too high on an IQ test. But the New London police department also turned applicants away for scoring too low on the same IQ test (and I assume that’s the main reason they used it). And it’s not very difficult to find more examples of companies using IQ tests, with or without some kind of minor tailoring to the job requirements. See, e.g., http://www.wonderlic.com/resources/client-stories
The Federal Reserve was created by an act of Congress, the salaries of the board members are set by legislation, the board is appointed by the President, it turns over its profits to the government, and can be audited by the GAO, within parameters set by Congress.
None of this is true with electric companies. Power companies are private, investor-owned organizations* that stepped in to provide electricity and, due to the realities of electric distribution, ended up in a monopoly position. Local governments negotiate – not set! – rates to (maybe) ensure customers don’t get gouged, and provide (or deny) approval of infrastructure projects, based on community feedback, but that’s essentially it.
Your state government doesn’t look at projected demand levels and instruct the power companies when and where to build new plants (though they do tend to have a hard-on for insisting on renewable energy installation), it doesn’t name board members or CEO’s, it doesn’t direct how day-to-day operations are run, it doesn’t get to audit the company, there really is no basis to call them “quasi-public abomonations” (to say nothing of the fact that this statement implies there’s a free market solution to electric distribution)
*excepting local electrical cooperatives, which generally build local infrastructure to connect to the power company’s grid.
Using aptitude testing to determine who in the company would be a good candidate for training for more “advanced” positions should:
a) Save the company money spent on training candidates who really don’t have the brainpower to do the job* and
b) Could allow the company to offer a lower salary, since the candidate 1) won’t have outstanding student loans to budget for, and 2) would likely expect less as they don’t otherwise have significant credentials.
As you mentioned, it’s good for employees for whom college isn’t a good option.
@Vox, on integrity: That depends on the field; the conventional wisdom wrt programmers is that a smart, lazy candidate is even better (code quality-wise) than a smart, industrious one. Programming in general tends to be an exercise in avoiding duplication of work.
If there really is a whole lot of money on the table, then I don’t believe that it’s the cost of producing a study and some legal uncertainty that’s keeping anyone from trying.
I don’t have access to westlaw anymore but maybe someone that does could take a quick peek and see if there are any cases in the lower courts.
@switchnode
This is one of those things that people say but don’t really mean. It is true that you want someone that’s going to look for opportunities to multiply his effectiveness through code rather than taking the straightforward but perhaps tedious route every time. But lazy really isn’t the right word, the lazy guy is going to be hanging out in the breakroom or posting on facebook not hacking a lisp macro processor into the jvm to avoid having to run thousands of lines of java.
I mean is Carmack lazy? Bellard? djb?
@ brad:
Exactly.
Think of the guy who “outsourced his own job to China“. In some sense, that’s the good kind of lazy. He, or at least his Chinese subcontractors, were getting work done.
But it was also a massive security vulnerability. That’s how he got detected. That’s why he got fired. Not to mention who knows if this was violating all kinds of labor laws and government regulations that could get the company into trouble.
That’s why employers don’t like the high ability / low integrity people: they’re the ones who are going to get you into hot water. Even the low ability / low integrity guy is better because he just hangs out in the breakroom instead of coming up with clever schemes.
Yes, yes, which is why I a) specified “code quality-wise” and b) stipulated that it was conventional wisdom and not gospel. But I think there is still a kernel of truth. Two remarks (as long as we’re wandering off topic):
– I was using ‘laziness’ to mean avoidance of tedium, not effort per se; the lottery of fascinations comes into play here. It makes writing the Lisp preferable to writing the Java, but it also (causally or correlationally) makes it more likely that writing the Lisp is preferable to scrolling Facebook.
– Carmack is an interesting example! He got picked up as a teenager for breaking into a school building using homemade napalm to steal computers—not the conventional idea of integrity. That’s extreme, of course (and Carmack learned his lesson), but look at hackers’ history with lockpicking, vadding, phreaking, ‘anarchist’ textfiles, and piracy; impulses towards rulebreaking, especially through collection of illicit knowledge, are incredibly common.
(The selection bias goes the other way, but they’re mostly benign, because the people involved are mostly bright enough to know better. I’m pretty sure all of my programmer friends found out how to make thermite as teens; I know none of them ever actually tried it. Does staying out of trouble count as “integrity”, or only honest virtue?)
Possibly when people say ‘smart but sneaky > smart and stolid’ what they actually mean is ‘smart but sneaky; smart given stolid is statistically unlikely’. And possibly this is a massive subcultural bias, based on contingent rather than convergent archetypes. But that is an anecdote fight for another thread, lol.
Anyway, again, this is re: code quality. It’s not everything, but you do really need some of it—break room guy won’t give you any trouble, but if all your people are break room guy you have no product.
Property developers know perfectly well that black people aren’t willing to pay extra to have their house on a golf course. Hence the proliferation of higher-end suburban subdivisions with included golf courses.
The obvious example is health insurance.
I mean, I don’t see this as a problem. But many people do.
In any case, such discrimination has already been made illegal by the Genetic Information Nondiscrimination Act, signed by George W. Bush in 2008. It’s almost exactly the “wow, Gattaca is a really scary scenario Act”.
The only person to vote against it was Ron Paul.
— Hey, do you think Vox or Schmox should be promoted?
— We’ll promote Schmox, of course.
— Uh, why?
— I did a bit of research, and it turns out Vox’s aunt had gyrotropilosis.
— So?
— It’s genetic! Vox might have the gene, and if he gets sick in the next few years, I’d rather have to replace a lower-level employee.
Doesn’t look much like Gattaca to me.
Is the part where you say it doesn’t look like Gattaca supposed to be sarcastic? It clearly does look like Gattaca, but I don’t think there’s anything wrong with it.
There is nothing wrong with this scenario. If genetic discrimination is used against everyone, we are not, on average, any worse off. We are better-off because people make more efficient decisions on the basis of superior information.
I benefit all the time from things, such as my intelligence, that I have no control over. I don’t claim moral credit for being intelligent. But it would be pretty stupid for someone to demand that my college had treated me with a near-perfect score on my admissions tests equally with a person with terrible scores. That would be socially inefficient and contrary to the purpose of the university.
Now, if I had this disorder and were discriminated against in turn, it would only be fair. Perhaps “fair” is not the right word; it’s not cosmically fair; it’s not a “level playing field”. But it’s the appropriate decision.
My half-brother had a stroke when he was born that impaired his brain function. It’s certainly not fair in a cosmic sense that this happened. He and my father can rage against the heavens. But if some company is determining whom to hire, they are obviously entitled to take that into account. That doesn’t mean he can’t get a job. It just means that, all else equal, he can’t get as good of a job as someone who is not impaired in that way.
If a genetic test reveals that someone has a 100% chance of getting cancer by age 30, it is certainly not cosmically fair that this person has to pay for the treatment. He or she did nothing to deserve it. But it’s not fair—or, more to the point, just—to make other people pay for it, either. I don’t think other people should be forced to pay for it, either directly or indirectly by lumping them all in together for health insurance and paying a “group rate”.
I wonder if the Reds are more okay with cosmic injustice (people being penalized for things over which they had no conceivable control) in this world because they are assured God will make it right in the next. And I also tend to think, of course, that for many leftists, the state takes the place of God. It can’t be a coincidence that communist dictatorships demand atheism.
Today, discrimination laws are enforced on the idea that one should not discriminate on qualities that don’t affect your ability to do the job: you can’t hire flight attendants on the basis of hotness anymore, but you can hire only hot strippers. The scary thing about the genetic info is precisely all the relevant info it threatens to provide.
@ Vox Imperatoris
In Gattaca, genetic science and engineering were at a very different stage and played a different role in society.
So, it doesn’t bother you that some people would be prevented from realizing their full potential because they might have a gene that might cause some illness at some point?
The imaginary Vox in my example will live the rest of his life in perfect health, wondering whether something was wrong with all of his work or with every employer’s judgment.
Meanwhile, poor Schmox will start drinking on the job at the age of 28. He has a new mutation that makes alcohol very addictive, and there’s no genetic test for it.
@ Nita:
People who have a gene that might cause illness at some point have less expected potential than people who are exactly the same but don’t have that gene.
If the odds are small, then the negative effects will only be small. If the odds are large, then the negative effects will be large.
Suppose Vox has a 99% chance of getting the disease next week. Then should the company still treat them equally?
But suppose Vox has a 1% chance of getting the disease in his lifetime. Then do you think that Schmox will have three or four times the job prospects? On what possible basis? And remember we’re assuming all else equal; in real life, it’s not.
Some people have degrees from Harvard. Others have degrees from Columbia. And still others have them from Auburn. The ones from Harvard get the best jobs, the ones from Columbia get just a little worse, and the ones from Auburn get a good deal worse, but they all can get jobs.
On what basis can the guy with a degree from Auburn complain that he’s not being treated the same as the guy from Harvard? The guy with a Harvard degree has a greater expected value. It doesn’t mean nobody from Auburn ever turned out to be a better worker than a guy from Harvard; but on average it’s not the case or the reputation of the two universities would be reversed.
(And since you’re probably not familar: Auburn University is not infamous or bad or anything. It’s just average.)
Does he not have access to the tests? Do the employers not tell him? In Gattaca, I suppose the employers don’t tell people because discrimination is technically illegal. Classic unintended consequences.
If a paraplegic wants to be a beat cop and doesn’t get the job, does he have to be a genius to figure out why? What is the moral difference here? He doesn’t get the job because he’s less qualified.
Come on, now. This is about probabilities. It’s about expected value.
Some people actually win the lottery. That doesn’t make it a good investment. You can win a hand of poker with unsuited 2,3; that doesn’t mean you should go all in with that every time. You can lose with pocket aces and often will; that doesn’t mean you should fold them.
The smart money was on Schmox. Sometimes bets don’t work out. Maybe they hire Vox and he gets hit by a bus.
It reminds me of a skit on Family Guy. Peter Griffin wins a contest and has to choose between two prizes: a boat and a “mystery box”. He says: “A boat’s a boat, but the box could be anything! It could even be a boat!” He opens the box and it’s some kind of worthless thing.
With the alcoholic-Schmox example, you’re saying employers are going to be better off if they take the mystery box. Or really, not even that: that they are going to be better off taking a boat that for all the world appears to be slightly inferior. Why? It makes no sense.
@ onyomi:
Well, if we’re going to get into uncharitable, psychologizing explanations, here’s what I think (in addition to what you say, which I tend to agree with):
It’s got something to do with your attitude toward hierarchy, toward superiority and inferiority. This is kinda sorta related to “respect for authority” but not totally.
I think that socialists, progressives, whatever vague label you want to use, tend to have the attitude that nobody can be better than anybody else. For instance, I was reading on sine-salvatorem‘s Tumblr a little while ago, this remarkable passage:
Now, my thought was: what do men and women have to do with any of the rest of these things?! There’s nothing wrong with any of the rest of them. Bishops are superior to priests (granting the validity of the system), bosses to employees (for the most part), teachers to students (obviously! Why would he object to this one in particular?), parents to children (as a general rule, they are wiser!). But men aren’t superior to women, so I agree in rejecting that one.
I feel like progressives reject all hierarchy. Conservatives are way too aroused by it and way too eager to defend existing hierarchies.
But I feel like, I don’t know, I have a…balanced attitude here. There are legitimate reasons to respect and honor people who are wiser, abler, and just better than you. And there are equally strong reasons to reject people who illegitimately pretend to that authority and/or respect.
So, for instance, when I was growing up, I respected my parents. Not just because, or from some kind of “filial duty”, but because they seemed like they knew what they were doing. I had occasional disagreements, but they were really minor and usually I was in the wrong.
But just because I respect legitimate superiority doesn’t mean I feel any sort of deference to some Duke Rollingham or whatever. So I absolutely feel the “democratic” resentment against nobles and monarchs and find the (especially American) obsession with the British royal family absurd. They’re not better than me. They don’t have some kind of divine right.
And neither does Obama or Bush, given the divine sanction of The People. I guess conservatives don’t respect Obama either because they think he’s a heretic or a usurper. But definitely under Bush (and even to some extent under Obama), there is this idea of Respect the Majesty of the Office.
There’s a Russian joke:
To twist it in my own way [edited to add more categories and complete the joke]:
With progressives, all authority is illegitimate except the legitimate.
With socialists, all authority is illegitimate, even the legitimate.
With conservatives, all authority is legitimate except the illegitimate.
With reactionaries, all authority is legitimate, even the illegitimate.
With me, what’s legitimate is legitimate, and what’s illegitimate is illegitimate.
Now, of course I’m being totally uncharitable and speculating, and of course everyone thinks he himself has the right view of things.
But I feel like if you give the conservatives an inch, they take a mile. You concede that, yes, it’s possible for some people to be better than others and the next thing you know they’re at men are better than women and white people are the superior race. You talk about IQ, and the “race realists” come crawling out of the woodwork. You say that men are better on average at football than women, and suddenly feminism was all a mistake.
And with progressives, the minute you say someone can be better than others, it’s because you’re trying to justify your privilege and your own unjustified superiority. It’s like “I can’t imagine anyone being better than me, so if someone else says certain people can be better than others, it’s because he thinks he’s better than everyone else. He couldn’t possibly think anyone’s better than him.”
Anyway, that’s my stream of consciousness on the subject.
***
Also, I can’t believe I left this one out:
Conservatives, or perhaps more appropriately reactionaries, believe in inherent or innate moral superiority.
Progressives believe in unconditional moral equality that can’t be taken away by anything you do (except when they’re waging holy war, I guess).
I believe in innate moral equality that diverges once people start making choices.
***
And just as a final reminder: I am well aware that not all progressives or conservatives share the views that I have attributed to them as a group. I’m really just thinking out loud here.
>The order, by rank and status, was as follows:
>….
>– Men before women
Etiquette guides say the opposite: women before men.
http://emilypost.com/advice/making-introductions/
“FIRST: Your grandparents, parents, or anyone older than you…. THEN: Your contemporary (or younger)
“FIRST: Your friend…. THEN: Another family member
“FIRST: An adult…. THEN: A child
“FIRST: A woman…. THEN: A man
“FIRST: Someone with a title: Senator, Mayor, Judge, Colonel, nobility, Bishop, Reverend, Professor, Doctor; anyone senior in rank to you (boss, CEO)…. THEN: Your contemporary (or younger)”
Update: here’s a version from 1922, in case you were thinking this was a new policy:
http://www.bartleby.com/95/2.html
“The younger person is always presented to the older or more distinguished, but a gentleman is always presented to a lady, even though he is an old gentleman of great distinction and the lady a mere slip of a girl.
“No lady is ever, except to the President of the United States, a cardinal, or a reigning sovereign, presented to a man.”
To avoid confusion, it is important to notice the difference between (a) which side to introduce first, and (b) which person within the same side to introduce first.
For example, if we have a higher-status person A, and a lower-status person B, this is how we would introduce them to each other:
“A, this is B.”
“B, this is A.”
But this is how you would introduce them as a team to another side:
“X, this is A, and this is B.”
It may be somewhat confusing that in the first example “this is B” precedes “this is A”, while in the second example “this is A” precedes “this is B”. If I understand it correctly, the reasoning is the following:
The highest honor is to have other people introduced to you. The second highest honor is to be introduced to other people. The least honorable is to be ignored. We try to give the highest honor to the people with highest status.
In the first situation, the highest honor (having other people introduced to you) is given to A by saying “A, this is B”. In the second situation, this option was given to X, so the highest remaining honor (being introduced) is given to A, while B is ignored for the few seconds.
Then it gets complicated if you have multiple people on each side, or if in given situation some rules contradict other rules (e.g. a younger woman vs an older man, not relatives). I don’t understand that too much, but I guess that rules based on more permanent traits are given priority, e.g. gender trumps rank. But it may also depend on context, for example the rank will be more relevant when the situation is related to the job.
I was thinking conscientousness/willpower/long term day-to-day performance, not IQ. No reliable quick tests for that.
Now we have college to prove these things, which is, of course, ~wasteful signalling~, blah blah, but can be at least masked and offset. Genetic testing would throw these differences into a painfully sharp relief. And oh yes, I’m thinking of myself here too.
Good point, but let’s remember it works both ways. If I don’t have a portfolio or any track record of “creative” output but my genome suggests I have an above average potential for creative expression, I may qualify for/get a job that I would have otherwise never had a chance to even interview for.
Thinking of myself as well.
Here’s a fascinating story about two pairs of identical twins accidentally shuffled at birth and raised in rather different environments: http://www.nytimes.com/2015/07/12/magazine/the-mixed-up-brothers-of-bogota.html
To help readers keep straight who is who in that fascinating article, I made up a helpful illustrated table:
http://www.unz.com/isteve/nature-vs-nurture-two-pairs-of-identical-twins-switched-at-birth/
Good write-up, I was thinking of doing one myself. Instead, I wrote an expansion on your post.
Some numbers on non-shared ‘environment’ effects
Scott Alexander has a new post out summarizing the interpretations of what constitutes non-shared environment in the ACE model estimates from standard (MZ-DZ) twin studies. I’m happy that he wrote that up because I had been thinking of writing up the same points, but he is a better writer than I am. Still, his post is a bit short on numbers and supporting research, so I will attempt to address that by presenting some supporting data in this post.
Entire post http://emilkirkegaard.dk/en/?p=5852
So what Turkheimer seems to be saying is non-shared environment is still important, but there is little evidence for *any specific part of it* to be reliably important in itself for all humans?
Which makes sense. Like a pie chart with a few thin slices labelled “exposure to lead” or “domestic abuse”, and one huge slice labelled “everything else”.
Epigenetics is strictly speaking just a molecular mechanism for the others. It gets a lot of silly press coverage, but all it does is tell us how things like biological insults, biological noise (you get epigenetic drink just like genetic drift) upbringing etc act in the body. The existence of epigenetics doesn’t really change our predictions about how heritability works.
(With the exception of epigenetics which can be passed from generation to generation, but it’s not clear if this happens much in humans)
Instead of “nature/nurture” I like to think of “nature/nurture/neither” with lots of the “non-shared environment” stuff coming in the “neither” column. “Nuture” sounds systematic and deliberate; lots of non-shared stuff is noise. Perhaps “nature/nurture/noise”. Either way, there should be three Ns and not two.
It makes scientists unhappy to not track these things down.
This short lecture by Kevin Mitchell is really great in making the point that much of the non-shared environment is “developmental noise” that is inherent and unavoidable. I especially like the facial (a)symmetry example.
Thanks. The bit about sensitivity to noise in the environment also being genetic (mostly? there might be a noise component there too) was fascinating.
I’m trying to figure out whether facial asymmetry is the only plausible surrogate for insensitivity to noise.
I hope this isn’t too tendentious, but I’m very curious about the moral implications of IQ.
Let’s take these propositions:
1- IQ is largely due to genetics
2- The main non-genetic factors that influence IQ (exposure to lead, traumatic stress as a child) are also impossible for someone to alter for themselves
3- IQ has a large influence on material success (income, etc)
In my experience, conservatives are far more likely to believe those 3 points than liberals. However, if you accept these points as true, it seems to be completely counter-intuitive to praise the success of wealthy people as a result of their own hard work and to hold them up as exemplars.
Murray, for example, is a libertarian who really strongly believes in IQ – yet his solution to the plight of people who are left behind by the knowledge economy is to argue for a sort of civic religion… his solutions seem to be incredibly palliative.
Basically – if you believe that the IQ you are born with is random genetic luck, and that IQ has a strong influence on your success in life, how can you coherently argue against redistribution on the grounds of fairness without resorting to quasi-calvinist arguments?
In the American context? I’ll believe it. But then America has a specific ethos about anyone working hard enough being able to achieve success, which ethos is largely orthogonal to reality.
If you can do nothing but relieve the symptoms, because the illness is incurable, then why should palliative measures be wrong?
What do you mean by “quasi-calvinist” arguments?
I’d argue against redistribution on the grounds of “fairness” by asserting the lack of claim the would-be recipients of redistribution on the resources of the would-be victims of redistribution.
Actually, the more I think about it, how could one possibly argue FOR redistribution if resource acquisition is mostly genetically predetermined?
I hereby commit the crime of arguing by analogy. And the further crime of arguing for something I don’t even believe. But I think it answers the question of how one could “possibly argue FOR redistribution”.
Imagine if we lived in a world where the primary way to get wealth was to inherit it; just like the primary way for French nobility in the 1700s to get land was to inherit it. You could imagine a (suicidal) peasant saying “all these nobles didn’t earn this land, they’re just lucky to have the right parents; we peasants have just as much right to it as they do!”
Likewise: “all these rich folk just got lucky to have the right parents (and hence genes); they didn’t choose their parents and neither did anyone else; how come they get to keep all the riches?!”
I don’t find this argument fully convincing (or else I’d be a die-hard communist). But at this point it’s your turn to explain how one could possibly argue AGAINST redistribution “if resource acquisition is mostly genetically predetermined”.
>“all these rich folk just got lucky to have the right parents (and hence genes); they didn’t choose their parents and neither did anyone else; how come they get to keep all the riches?!”
Because it is theirs. What other argument do you really need?
(I am presuming that someone up the line of inheritance, probably multiple someones, earned that wealth, and the dynasty managed, in their stewardship, not to lose it, and to bequeath it to the current generation.)
Well, if we redistribute wealth to the people it will become theirs and no other argument will be needed.
In a very rough simplification, if we trace back aristocratic wealth up to the roots of feudalism then “earning” it consisted of declaring an existing resource (land) to be yours and being strong enough to beat up everybody who disagreed.
>Well, if we redistribute wealth to the people it will become theirs and no other argument will be needed.
And why should the owners of the wealth consent? Redistribution is literally taking their stuff and giving it to someone else. If you object to right of conquest, why do you not object to redistribution?
Usually, the argument against redistribution is made on the basis of fairness: we shouldn’t take money away from people that worked hard for it to give it to people that did nothing to deserve it.
If you believe that the ability to earn money is innate and the result of luck, then that argument falls apart. The reason I called it a ‘calvinist argument’ is because if you believe that people intrinsically deserve what they are born with, then the people who are born just deserve it in a cosmic sense.
>Usually, the argument against redistribution is made on the basis of fairness: we shouldn’t take money away from people that worked hard for it to give it to people that did nothing to deserve it.
That’s actually something I’ve not yet encountered. The argument against redistribution I’m familiar with is that it is legalized theft. Unless the current owners acquired their wealth illicitly, they are entitled to keep it.
This is a really strange, somewhat contradictory position.
“Legalized theft” is either an oxymoron, or “theft” indicates something that is not merely illegal but immoral — you are trying to argue that redistribution is immoral.
But then you justify any other means of wealth acquisition by saying it’s OK as long as it is not illicit which just literally means not illegal.
The upshot is that all you have done here is underscored your assertion that redistribution is immoral. You haven’t made an argument for why it is immoral; you seem to think it is self-explanatory (“legal theft”). And you haven’t addressed the point at hand, which is that if wealth acquisition is genetic, then it is not subject to moral judgment since individuals do not choose their own genes — no one “deserves” their level of wealth because it is not dependent on choices made by the individual.
Yes.
I chose “illicit” instead of “illegal” consciously, in order to refer to natural law, rather than potentially arbitrary legislation of an arbitrary polity. If, say, Turkey would make shooting Kurds on sight legal, it would not make shooting Kurds on sight any less murder than it was before the legislation made it legal.
Yes.
No, they don’t deserve their wealth. Neither do the recipients of redistribution deserve to have the wealth-holders’ wealth redistributed to them. But so long as they don’t steal it in some fashion, they are entitled to have what they have.
I don’t believe in natural law. If there is such a thing, it falls under the rubric of “immoral” rather than “illegal” which is exactly the distinction you are flouting to dance around the need to support your position with an argument.
Asserting that something is self-evidently immoral does not establish that it actually is immoral. Especially when you subsequently concede that it is not immoral under the premises which we are considering:
This is, essentially, an admission that redistribution is neither moral nor immoral — it simply does not have a moral valence under the assumption that wealth distribution is entirely genetically determined. But this clearly contradicts the assertion that redistribution is immoral (even if you use the term “illicit” instead).
From what I can gather, you are not making a coherent argument against the sentiment with which you disagree. Are you expressing your disagreement to signal your political affiliation instead?
@ Anonymous:
You are not answering the question of why they are entitled to their property. What is the basis of the entitlement?
I give one answer in another comment, but you’re just asserting obviously they’re entitled to it. Not good enough. If people have to ask why, it’s not so obvious.
You could say the same thing about murder. Why are people entitled to their lives? What would be wrong with a country taking away the entitlement and sending people to the gas chamber? Surely there are answers, but you have to actually give the answers.
@wysinwyg
This is true on a moral theory which makes rightful ownership a function of desert. But Anonymous seems to endorse a moral theory which makes rightful ownership a function of just acquisition, not desert. It is true that he has given no argument for this position, but I don’t see it as incumbent upon him to offer one– claims like “S should have unrestricted rights to some piece of property iff S justly acquired that property” and “S should have unrestricted rights to some piece of property iff S deserves that property” look awfully foundational, like they could only be supported by direct intuitions rather than syllogisms from further premises.
I guess?
I did not concede it – I don’t recall saying anything to the effect of “OK, you are right”.
Deserving something and having something are different things. If you say you deserve something, but don’t have it, and have no other claim to it, like having it taken forcibly from you, I’m not going to agree with you. Indeed, even if you do have it, I’m not going to say you deserve it, merely that you have it and can keep on having it, regardless of your deservingness or lack of it. If, as we are taking for granted here, wealth comes to people by means of luck, whether genetic or otherwise, they don’t deserve it any more than a lottery winner deserves his win. But we don’t propose that the lottery winner’s winnings should be redistributed to everyone who also bought a ticket and did not win, are we?
And what might that be?
Possession.
At a higher level of complexity, you could say: Possession through licit means.
I’m not sure I can go deeper down.
It’s against natural law.
Consider three people. One is an Actor, another is a Reactor, and the third is a Reasonable Observer.
The Actor performs an Action, to which the Reactor responds with visiting violence on the Actor. The Action is said to be breaking natural law if the Observer, being Reasonable, does not find the Reactor to subsequently be a threat to himself and his livelihood.
So if someone robs you and you beat them within an inch of their life, the onlookers are not likely to think you some kind of arbitrary batterist. On the contrary, if someone gives you a gift, and you beat them within an inch of their life, onlookers are going to be concerned whether you are going to do the same to them.
@ Anonymous:
This is like pulling teeth! What is the standard by which you judge which means are licit or not?
What is the standard of Reasonable with a capital R? How is it different from the regular kind of reasonable? What is the basis upon which they judge how to treat this criminal? Self-interest? The good of the criminal? The greatest good of the greatest number? Kant’s categorical imperative? Women’s intuition? WWJD?
One interpretation is that you’re saying, in a long-winded and vague manner, that we punish people who commit unprovoked battery because it’s in everyone else’s self-interest to do so. If so, I do not disagree.
I don’t disbelieve in “natural law”, if you mean that there are naturalistically-grounded reasons for preferring some courses of action and some political systems to others. But just saying “it’s true because of natural law” is uselessly vague. It doesn’t tell me what you think the natural law is, or on what basis we adjudicate disagreements as to its content.
Primarily on the basis of whether theft occurred or didn’t (I’m not sure if there are cases which don’t simplify to theft). I guess the question becomes whether theft is something that exists without a legal entity to define it and enforce punishments for it (which I believe it does, and wysinwyg implies not to).
I’m just labelling my variables. Reasonable in this context would mean something to the effect of being capable of predicting future events based on prior ones, and constructing mental models of other actors, which are both not dissociated from reality (doesn’t have to be super-accurate).
Self-interest. The others seem like “advanced morality” which is outside of the scope of basic natural law.
Right.
If I see someone redistributing someone else’s property on the basis of this someone’s arbitrary judgment, I’m going to feel concerned that I might be next.
@ Anonymous:
See, now we’re getting somewhere! Before, you were just making an emotionalistic type of argument, using words and phrases that have positive emotional valence among opponents of redistribution and negative valence among proponents. Making it a really good way of preaching to the choir.
It’s what Scott has called the noncentral fallacy or the “worst argument in the world”.
Leftists do not dispute that taxation for the purpose of redistribution is a type of “taking” (to use a legal term slightly out of context). They dispute the idea that it’s unjustified.
In the same way, proponents of capital punishment do not dispute that it is a type of killing. They dispute the idea that it’s murder, i.e. unjustified killing.
I have no problem with saying “taxation is theft!” as a slogan. But it’s just a slogan (unless you’re a moral intuitionist like Earthly Knight, I guess). It translates to: I think taxation is unjustified form of depriving people of their property. It means you think there is no morally relevant difference between the government doing it and a common criminal doing it, at least for the purpose of redistribution.
Leftists think there is a morally relevant difference: namely, it’s the will of the people, or it’s for the common good, or it promotes fairness and equality. It is, of course, much more orderly and predictable than theft by common criminals.
Most importantly, once you say that taking property from people is not wrong “just because” but because it’s against everyone’s self-interest, we have to consider whether it’s possible to have a situation where taking property from people is in everyone else’s self-interest. Or a situation where it’s in the interest of some of them but not others.
Karl Marx believed that the workers ought, or at least will, act in their class interest, which is basically the general self-interest of all of them. But he thought their class interest was opposed to that of the bourgeoisie, leading to inevitable class conflict. Therefore, the right thing for the workers to do was to seize the means of production, according to proletarian morality. While according to bourgeois morality, the capitalists were quite justified in using force to old onto them. He just thought the workers were destined to win.
Now it seems your real dispute with Marx is not whether people of any given class ought to do what is in their interest. You disagree on what their interest is. And this opens a door for us to appeal to the facts, to economic science, and decide. Does the existence of private property rights, even for the rich, benefit the poor? Or would it be more in their interest to see property rights restricted to some extent, or to have some but not all property taken away and redistributed?
I don’t agree with all of David Friedman‘s metaethical theorizing. But makes the following observation, which is very important:
Similarly, in the dispute between you and Marx, it seems you disagree mainly on the economic facts, not the ethical standards.
I am largely convinced by Friedman’s thesis that many, if not most seemingly ethical disagreements between different sides of the political spectrum in fact boil down to working (or wanting to work) with different sets of facts, rather than actually differing ethical standards.
I believe Friedman has used this to argue in favor of a fact-based approach over a moralistic approach to arguing for libertarianism: prove to the opponent that libertarianism objectively produces the best results and you don’t need to change their moral framework, since they, like you, ultimately just want people to be happy.
But, as indicated in the quote above, differing beliefs push people to gravitate towards certain facts. If you are already inclined to believe that capitalists are ruthless exploiters then you can find lots of historical examples of that. If you are already inclined to believe that capitalists are hard-working people looking to make a profit in the process of meeting their customers’ needs then you can find lots of historical examples of that as well. If I say to the person predisposed to believe capitalists are ruthless exploiters, “well, just posit, for the sake of argument, a kind, hard-working, honest capitalist…” then they can just as easily respond, “okay, but most capitalists irl aren’t like that.”
Similarly, even if I say to the socialist: “look at what a failure the USSR, the DPRK, and Maoist China were,” they can say “but those weren’t real socialism, just fascist authoritarianism in the guise of socialism. Look instead at Sweden,” and if the socialist says to me “you want anarchy, well I guess you’d like to live in Somalia then?” I say “well, Somalia was better off without a government than Somalia with a government or its neighbors with governments, even though it’s still not great in an absolute sense. Look instead at Singapore, Hong Kong, South Korea, West Germany…”
Obviously I have my personal opinion of which set of facts is more persuasive, but it also happens to be one which accords with my moral sense that the free market mostly works pretty well and is ethically justifiable as compared to using coercion to tweak it. If I had a different underlying moral sense I might gravitate towards a different set of facts.
Since you can find facts to support almost any ethical intuition, maybe ethical intuition is, in fact, the primary cause of disagreement, even though you can get everyone to agree on “yes, you should save that drowning child.”
@ onyomi:
The first thing I would say is: I don’t believe in some kind of innate moral intuition, certainly not about high-level matters like capitalism vs. socialism. As Richard Lawrence says in a criticism of Michael Huemer:
I think that description of the nature of “moral intuition” is exactly right. And a lot of political rhetoric is based on pretending (honestly or dishonestly) that it isn’t so. For instance, I could rephrase the Declaration of Independence:
Based on our familiarity with similar situations, we have an inarticulate sense that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness, etc., etc.
Now, that’s not quite fair. John Locke articulated their theory quite well, though not perfectly. But none of it is genuinely self-evident.
***
That being said, I think there is some kind of important psychological difference between say, David Friedman and Murray Rothbard, or David Kelley and Leonard Peikoff. Friedman and Rothbard are/were both anarcho-capitalists, with hardly any daylight between them in terms of practical policy positions. Kelley and Peikoff are both Objectivists with hardly any daylight between them in practical policy positions. (And there’s hardly any daylight between minarchists like the latter two and anarcho-capitalists like the former two.)
Yet Rothbard had basically the temperament of someone like Trotsky, while Friedman has the temperament of someone like Ezra Klein. And Rothbard hated Friedman, though Friedman did not hate him back except insofar as he resented his hostility.
The point is, I think, that basically any political philosophy (or moral philosophy) can be framed in such a way that it is persuasive to people of any temperament. This is the case regardless of whether one’s temperament is learned or innate or a mix of both.
Even Nazism is like this. Obviously, as framed by Hitler, it is designed to appealed to people of a certain temperament or “intuition” such that they hate and resent people who are not like them. It doesn’t tell you to see things from the Jewish point of view or even have a hint of sympathy for them. So since Hitler framed it like that, it’s going to appeal more to certain type of person.
But, at the same time, Nazism has arguments for people who don’t feel drawn that way. For instance, just look at Himmler’s infamous secret admission of the Holocaust:
What is he saying here? That “some, I assume, are good people,” (okay, that was a cheap shot). He’s saying that, yeah, we tell people that the Jews are all evil, but some of them really are decent or at least seem to be that way. Nevertheless (according to his twisted view of the facts), the greater good means that they have to be dealt with. But it’s not an easy thing, and he understands that it’s hard, just like it’s hard to shoot “comrades” who have their sympathetic stories but are undermining the Party.
(And this is actual Himmler, not the hypothetical “bleeding-heart Himmler” I can imagine.)
That’s Nazism for the “liberal”, “bleeding-heart” mentality: what sort of kindness or decency is it to leave these Jews alive so that they can destroy the whole world and bring us under Bolshevism and tyranny? That’s the real callousness. If you really care about the poor, the weak, the downtrodden, you’ll do everything necessary to keep this from happening.
(People looking to quote me out of context are going to have a field day with this one! For what it’s worth: I absolutely deplore and reject Nazism, and any form of racism, collectivism, and totalitarianism, as should be clear from everything else I have written.)
***
All this is to say: if there are genetic differences in temperament, I think they have almost nothing to do with any concrete political opinion. It all depends on how these temperaments are socially guided into backing different positions.
Carry the argument a few steps further:
I didn’t deserve to be born a human instead of a dog, so we should redistribute to give dogs the same level of utility as humans. He didn’t deserve to be a dog instead of a rock, so … (hard to measure utility of a rock).
He didn’t deserve to be born in Nazi Germany and become a concentration camp guard and kill people, so we shouldn’t blame him.
We are back with the problem of moral luck.
My response is that we don’t condition praise, blame, desert on the spirit before birth, the unembodied person, but on the actual person as he now is. If you end up as an honest, generous, productive person you deserve praise for that, if you end up as a dishonest SOB who gets his pleasure out of beating up strangers you deserve blame for that. Praise and blame are not being applied to you at birth but to the person you now are.
If one believes that people are entitled to control their own bodies, subject to not harming others, that people are entitled to own that which they produce, the fact that one person controls a body that has skills others value a lot or can produce things of value isn’t made irrelevant by the fact that the person at birth, or before birth, didn’t deserve to end up that way.
1. You seem to assume that we should apply praise and blame. Should we?
2. People not only do things, but also have things done to them by others. E.g., some businesses work by luring the easily persuaded into bad deals. Are reactions like “lol, if you’re dumb enough to fall for that, you deserve it” correct, by your reckoning?
3. IQ is not a skill of the body that “you” control. It’s not like the processing power of your computer. It is your own processing power, which affects your ability to control anything you might want to control.
@ David Friedman:
Is that you, Peter Singer?
Anyway, that seems to be the basis of the whole utilitarian drive for animal rights / animal welfare. Or rather, that there’s no morally relevant difference between being a human and being a dog.
If a given person born in Germany in the 1910s was inexorably determined to join the SS and guard a concentration camp, then I say we absolutely have no cause to blame him.
Never mind inexorable determination: if joining the SS rationally seemed like the best choice at the time, we shouldn’t blame him. I think it’s perfectly possible that there were honest people in the SS, who “evaded reality” to as little extent as you or I, but who were simply convinced by all the propaganda. Maybe it’s not likely; maybe there weren’t many of them; maybe there weren’t any; but it’s possible.
That doesn’t mean that their victims didn’t also have the right to fight back, but it is certainly possible in war for the men on both sides to be largely innocent.
This seems like a way of framing the issue so as to dodge or miss the essential question.
Pace Rawls, of course we’re judging the person as he is now, not as he was at birth. The question is: take the specific action the person is doing now. Maybe he’s a CEO running a successful company. Maybe he’s Jack the Ripper murdering prostitutes.
Does the CEO deserve credit? Does Jack the Ripper deserve blame? Well, we have to ask, how they get into these positions? What were the causes? If we trace all the causes back and it turns out that all of them go back to things outside of those individuals’ control, that everything apparently “in their control” was the inexorable result of those causes, it seems preposterous to award them either credit or blame. They’re just puppets, moved by larger and more basic forces.
From the Roman writer Lucian:
“You seem to assume that we should apply praise and blame. Should we?”
You seem to assume that we can apply it or not as we please. If so, why can’t we apply our talents as we please, and so deserve praise or blame?
Well, currently we do apply it whenever we feel like it, but that’s hardly optimal. I think we should try to understand whether or when it makes sense, improves the situation and is necessary, and then decide what to do.
“I didn’t deserve to be born a human instead of a dog, so we should redistribute to give dogs the same level of utility as humans. ”
On the other hand redistriburion doesn’t have to be redistribution to the point of complete equality.
There are many stupid arguments put forth by all sorts of people. Smart opponents of wealth/income redistribution don’t argue on the basis of fairness. This includes Murray, who argues in “Losing Ground” that welfare programmes are on net harmful to those they are supposed to help.
“Murray, for example, is a libertarian who really strongly believes in IQ – yet his solution to the plight of people who are left behind by the knowledge economy is to argue for a sort of civic religion… his solutions seem to be incredibly palliative.”
I thought his solution was basic income – which is also sort of palliative, but palliative care is exactly the kind of care you give when you think a disease is incurable with current technology.
He argued that wealthy, intelligent people should try to take up the hobbies of blue collar workers and segregate themselves less.
There was an exceptionally good take down of his last book by David Frum here: http://www.thedailybeast.com/articles/2012/02/06/charles-murray-book-review.html – it’s long, even by slatestarcodex standards.
I’m sort of confused as to why David Frum considers himself a conservative. His solution to the problems seems to be moderate social democracy and redistribution.
I haven’t read the review, but I trust David Frum less than I trust Charles Murray—and I’m not really a fan of Murray’s. For the basis of my view of Frum:
http://m.washingtontimes.com/news/2010/mar/18/a-tale-of-two-great-depressions/
or (same story, different version)
http://daviddfriedman.blogspot.com/2011/04/nicholas-kristof-gets-his-history.html
Those positions are more in keeping with traditional conservatism, such as that espoused by GK Chesterton, than what passes for conservatism these days. It only feels “off” when you compare it to the more libertarian strain of thought that passes itself off as “conservatism,” yet seems interested in conserving nothing – see Kasich at a Republican debate insisting conservatism stands for “growing the economy and producing jobs!” (and nothing else), or the recent meltdowns of NRO-types like Kevin Williamson and Rick Wilson
@ nyccine:
Yes, maybe Frum is 19th-century Tory type of conservative, in the model of an unprincipled interventionist like Disraeli.
Margaret Thatcher had some interesting things to say about this, as Wikipedia relates:
The thing is (to say nothing of Thatcher), normal American conservatism is a variety of liberalism—because for the most part, what they are trying to conserve is the liberal legacy of the American Revolution.
The Straussian neoconservatives are pulling a fast one because they are trying to conserve something very different. C. Bradley Thompson has some interesting things to say about it in this exchange in Cato Unbound between him and Straussian neoconservative Patrick Deneen. And Deneen nicely reveals the dishonesty of his rhetorical style.
Thanks for the Cato Unbound link. It’s a nifty introduction for those who want to play a few rounds of “Spot the Neocon” this election season.
@ tcd:
I should say that while Deneen exudes neoconservatives’ characteristic slipperiness, C. Bradley Thompson is really over the top in his hostility in response. It doesn’t help, and it doesn’t make him seem more persuasive.
Oh I agree, Thompson is excessively hostile to Deneen in his replies.
“wealthy, intelligent people should try to take up the hobbies of blue collar workers ”
Since there’s no reason to engage in a hobby if you don’t like it, and no obstacle to their doing so now — why on earth should they take up hobbies that they don’t like? At the very least, it raises the chance of making people more hostile to the blue-collar workers.
Because it rolls the upper- and lower-classes together at least a little. Football fans might then mock Patriot fans as rich toffs, but that’s a lot more of a shared existence than knowing that actual rich people wouldn’t be caught dead watching football.
There has been a vast change here. My dad was a lower class kid growing up in rural michigan in the ’60s. He played tennis. Tennis was a common thing back in those days. Now, it’s almost exclusively played by the well-off, and mostly the white well-off. Which makes it uncool, which means lower class kids will refuse to play it even if given the opportunity. Which reinforces the social roles that rich kids play tennis, golf, and swim, while poor kids play basketball and football.
For the rich and the poor to have some shared connection could be important for the cohesion of a nation. Murray certainly thinks so. I’m less sure.
“Because it rolls the upper- and lower-classes together at least a little.”
What if anything does that mean?
What benefit does society have from my thinking that the blue-collar workers liking some moronic game is why I have sacrifice hours of my precious leisure time engaging in the stupid activity of watching the dumb thing? Do you think that “shared connection” is fine even if it’s contempt and resentment?
And if people are using them as class-markers, you are chasing a rainbow: it will retreat as you advance. They can create more.
None of this is contrary to libertarianism or the idea that wealthy people, by and large, deserve their wealth.
It is perfectly possible to reconcile the following positions:
a) Metaphysical libertarianism: human behavior is governed to some extent by free will, as opposed to prior causes.
b) Common-sense semi-determination: human behavior is influenced to some extent by outside causes. There is obvious variance in innate ability and environmental influence (the two are not morally different, really).
c) “Anti-Calvinism”: people do not morally deserve rewards and punishments that are not the result of free choices.
d) Property rights: people who produce wealth are entitled to keep and transfer it without having it redistributed.
e) Respect for the hard work: we should respect people who work hard, even if they are of limited ability.
f) Superior respect for “Great Men”: we should nevertheless have higher regard for people who both work hard and achieve great things.
Examples of people who hold all these positions: Ayn Rand, Bryan Caplan and Michael Huemer (at least on most of the points for the latter two).
First of all, you have to recognize that the morally relevant question is not: “nature or nurture?” Nature and nurture are both forms of determinism. If “anti-Calvinism” is true, then no one can deserve anything either on the basis of the genes he didn’t choose to be born with or on the basis of the upbringing and society he didn’t choose to be raised in.
The next part of the puzzle is to recognize that property rights are, for the most part, based on what Robert Nozick called “entitlement”, not desert. This is obvious in the case of inherited wealth. What has the five-year-old heir to a billion dollars done to deserve it? Clearly, nothing. But if his parents were legitimately entitled to that money, they are entitled to pass it on to the child, and no one else has a superior claim on it. No one deserves that fortune; the fact that someone, somewhere may be in need does not go to show that that person has earned it.
What is the basis of the entitlement? That it is in the interest of everyone to accept a legal regime in which property rights are secure such that capital can be accumulated and transferred across generations without being broken up.
Nevertheless, we can also say that to the extent people exercise free will to work hard and earn money, they deserve to keep it; they have a desert-based claim as well as an entitlement claim. If it is true that the wealthy are mostly productive and virtuous, then they deserve the extent of their wealth that is attributable to their choice to be morally good. However, they are still entitled to all of it. And even [insert spoiled heiress of the day here] is entitled to her wealth, despite the fact that she hasn’t done anything to deserve it.
It is also true that the poor and middle class and entitled to their wealth, and they deserve the extent of it that is determined by their free choice. Of course, if they produce less with their work, they are entitled to less. But from a moral point of view, the poor person of limited ability is equal to the rich person of great ability, if they both exercise their potential to the fullest. So we should admire hard work in all people.
Nevertheless, there is something special about the John Galts of the world, as opposed to the Eddie Willerses. Even if they are equal from the standpoint of subjective moral virtue, the John Galts produce more objective good. Therefore, even though they may be equally virtuous, we should give superior respect and honor to the great people who do great things. This is out of gratitude and in the hopes that we way encourage ourselves and others to strive for more and be more like them.
Anyway, that’s the answer for how people can put all these “crazy ideas” together. I should point out that many political libertarians don’t believe in metaphysical libertarianism (the two have little to do with one another): they are compatibilists or hard determinists who simply follow the entitlement story only.
That’s a pretty good rationale. You can also add to that, which you lightly touched on, that property rights creates an incentive structure which contributes to those able to producing innovation to do so, improving the lives of those that aren’t by making luxuries and necessities cheaper or of better quality.
(edit: As is pointed out in the next comment down, of course.)
Yes, I was really sparse on the detail there. Being only slightly facetious, I could say that the whole science of economics (including public choice theory) is the reason why people ought to be entitled to their property.
In that context, I was mainly thinking about why people ought to be entitled to their inherited wealth. And of course it has to do with incentives for the people who are going to bequeath the wealth: if you know that the government will take all your wealth when you die, you’d better spend it while you can rather than investing it.
” I was really sparse on the detail there. Being only slightly facetious, I could say that the whole science of economics (including public choice theory) is the reason why people ought to be entitled to their property.”
And they are, for some values of “property” and “entititled”, in almost every system near-fascism to near-communism.
“In that context, I was mainly thinking about why people ought to be entitled to their inherited wealth. And of course it has to do with incentives for the people who are going to bequeath the wealth: if you know that the government will take all your wealth when you die, you’d better spend it while you can rather than investing it.”
I haven’t noticed bonfires of wealth in countries that have inheritance tax. But inheritance tax tends to be less than 100%, and that kind of tax doesn’t remove incentives, any more than typical income tax does.
A possible angle of attack is that there are different possible “legal regimes with secure property rights”, so the opponent is only attacking this specific regime (and the people who owe their wealth only to this specific regime), not the idea of such a regime in general (or the wealthy people in general).
Yes, of course that’s a potentially valid argument.
The usual response to arguments like “the arrangement of property in England is unjust because it goes back to the Norman Conquest” is that a) the people currently alive didn’t perpetrate that injustice and b) the deadweight loss and incentive effects from redistributing everything would far outweigh any gains to justice or fairness.
But there are other situations where that might not be so. For instance, the economist George Reisman, writing around the time of the fall of the Soviet Union, pointed out that it would be a terrible idea to simply turn the Communist Party into a private corporation and retain them as the owners of all the factories in the country.
On the other hand, even though Russia was in fact privatized in a very corrupt way, it would most likely be a very bad idea to “redo” it. Even the suggestion that this is possible is going to vastly reduce the incentive to invest long-term.
Or increase the incentive of the “oligarchs” to take all their wealth out of the country, i.e. divestment from Russia. Which indeed happened to a large extent because of fear of a Communist “redo” and/or fear of being on Putin’s bad side like Khodorkovsky.
“The usual response to arguments like “the arrangement of property in England is unjust because it goes back to the Norman Conquest” is that a) the people currently alive didn’t perpetrate that injustice and b) the deadweight loss and incentive effects from redistributing everything would far outweigh any gains to justice or fairness.”
But that is a response to an argument against complete once-and-for-all redistribution. It is not an argument against typical redistributive taxation (welfare state), and therefore not an argument for some kind of anarcho capitalim, libertarianism, etc.
VI
“d) Property rights: people who produce wealth are entitled to keep and transfer it without having it redistributed.”
And do *you* have access to a set of clear facts about what production consists of?
“That is the basis of the entitlement? That it is in the interest of everyone to accept a legal regime in which property rights are secure such that capital can be accumulated and transferred across generations without being broken up.
Nevertheless, we can also say that to the extent people exercise free will to work hard and earn money, they deserve to keep it; they have a desert-based claim as well as an entitlement claim. If it is true that the wealthy are mostly productive and virtuous, then they deserve the extent of their wealth that is attributable to their choice to be morally good. However, they are still entitled to all of”
Entitled to all of it, or entitled to all of it that they are entitled to? If property rights are decided by a legal regime, then it seems quite possible that the regime can decide that you are entitled to a percentage, and they are entitled to the remainder, to pay for the cost of of upholding your property rights, among other things.
What you suggest in the first paragraph doesn’t match what you suggest in the second.
There are two points here – an “effects” argument and a “deserts” argument.
Effects: A free market economy rewards pro-social behaviour and increases the size of the pie. If I set up a new business, the world is better off than if I didn’t. I am more likely to set up a new business if I get to keep the profits I make, than if a portion of the profits is confiscated from me and given to strangers. Therefore it’s better that I get to keep the profits. Even hard determinists believe people respond to incentives.
Deserts: Sure, there’s a sense in which I didn’t “earn” my IQ, my personality, etc, but that doesn’t mean they’re not mine, or that I’m not entitled to what flows from them. There is no me absent them! Heritability isn’t even the issue here – I’m sure you’d be asking the exact same questions if IQ and personality were environmental and malleable. No-one chose to be born to parents who played them Mozart in the womb or whatever. Your argument depends on the assumption that if causes external to me influence my actions, I am thereby absolved of moral responsibility (good or bad). Few agree.
It may be absurd to say that, if there is any degree of external influence, you are absolved of responsibility.
But we can split the factors governing the decision into the factors deriving from external influence and the factors deriving from free will. And you can therefore have greater or lesser responsibility depending on the balance.
That seems to be the whole point of “mitigating circumstances”. For instance, people say that someone who shoots the man in bed with his wife is guilty of a lesser crime because to a large degree he was “overcome by his emotions”. If the situation was enough to make him “temporarily insane”, he wasn’t guilty of a crime at all.
I’m not going to pretend that the American legal system has an entirely coherent outlook on moral philosophy, but there are definite ways in which ideas like this are commonly accepted.
Mitigating circumstances isn’t a claim that the defendant isn’t morally responsible for what he did, merely a plea that the situation in which the crime was committed was such that justice requires lenity. You plead mitigation after conviction – so you’ve already been found responsible when it comes your turn to convince the judge to go easy on you.
“Your honour, I shouldn’t have driven home from the pub that night, but I didn’t have any other way to get home, I was really stressed because my wife had just died, and I didn’t realise I was slightly over the limit. I’ve always been a careful driver and shown good character, and if you take away my license I’ll lose my job and cause hardship to my children. I know what I did was wrong and dangerous, and I’m sorry – please give me another chance.”
That’s how you plead mitigation. If you try to do so while disclaiming responsibility for what you did (“my wife had just died so I couldn’t help driving drunk”) you will aggravate the judge and wind up with a heavier sentence, which supports the view that mitigation isn’t about disclaiming responsibility, but rather that the judge should use his discretion in your favour, because what you did, while bad, wasn’t as bad as all that.
Diminished responsibility manslaughter (which is distinct from mitigating circumstances, at least in England and Wales), is normally seen as an anomaly caused by the fact that, historically, the death penalty was the only permissible sentence for murder, so dim res came in as a backdoor way of allowing a plea for mercy in cases where it wouldn’t be allowed.
More generally, I don’t think it’s useful to split decisions into “external influence” and “free will.” It requires an unnecessary commitment to metaphysical libertarianism, and nothing turns on it in popular theories of justice. I can’t find the link right now, but last year there was a large scale survey done by experimental philosophers, which showed that the most people thought that criminals were morally responsible for their crimes, even if those were entirely pre-determined by external causes. The philosophers naturally tried to use this to show that everyone’s secretly a compatibilist (ha!) but the point is that if criminals are morally responsible even in a fully deterministic world, then by the same token, entrepreneurs are morally worthy.
First of all, let’s get a definition in here.
Mitigating circumstances:
It’s not a claim that the defendant is completely morally innocent. It’s a claim that he’s not as guilty as the prototypical offender.
In the case you give, you say the defendant is “stressed” and “didn’t realize” he was over the limit. Those are, to some extent, factors outside his control. That diminishes his responsibility. But there are still steps he could have taken to avoid the situation, so he is somewhat responsible. Therefore, he deserves some punishment (or at least some blame and a reprimand) but not no punishment.
Saying “I couldn’t help it” is going to aggravate the judge because in fact he could help it. It’s not true. If the situation were: “somebody drugged me and tied me behind the wheel”, then it would be true. And in that case, the victim would not apologize to the judge and ask for leniency. He would just say: let me off, I didn’t do anything wrong.
(There are, of course, other factors the judge may consider, such as: how obsequious you are before him. I’m not saying it’s a perfect system.)
Mitigating circumstances is a partial form of an excuse, not a justification, that’s true enough. An excuse says that the act was still objectively wrong and that ideally you shouldn’t have done it, but in the circumstances you can’t be blamed (or blamed as much).
I don’t think it’s an unnecessary commitment, and I think it does matter.
For instance, if we’re not making a distinction between external influence and free will and wish to deter crime, we ought to punish people more when they’re under greater pressure or temptation to commit a crime—because the punishment should be strong enough to outweigh the provocation toward crime. If we make the distinction and wish to treat people as they deserve, we ought to punish them less when they’re under greater pressure or temptation.
Okay, but that’s bad because in a deterministic world, criminals would be morally innocent. Unjustly punishing them would not be made up for by unjustly honoring entrepreneurs.
Under determinism, criminals ought to be treated like patients in an asylum for the criminally insane. It would be absurd to blame the criminally insane for what they do, unless you redefine “blame” to mean Skinnerian negative reinforcement—in which case it would not be absurd but would be different from what is normally called “blame”.
These mass surveys are always finding crazy things, but one of the most interesting I’ve found is this one that talks about whether people’s actual decisions about just punishments are determined by “just deserts” or “deterrence”. Even though people endorse both (because they endorse basically every reason for punishment), they go almost 100% by “just deserts”. That’s not exactly the same as what we’re talking about, but it’s related.
Here is a closely related study with a similar outcome.
I agree wholeheartedly, but what bothers me even more than the metaphysics, is the moral mistake. A responsible decision is not one that comes out of nowhere. It is one that comes out of the decision maker, as they were at the time. It should reliably (not randomly) reflect their beliefs and values.
On the idea that the man who shoots the man he finds in bed with his wife gets off because of temporary circumstances …
In quite a lot of historical legal systems, he gets off because what he did was legal–including Periclean Athens, although shooting isn’t an option in that case. Which makes me suspect that the “temporary insanity” is a kludge, a legal fiction, a way of fitting our feelings about the situation to our legal theories.
Who are “we” here?
Sure, in some way the law is carrying out people’s feelings on the situation: the feeling that it’s not just to punish someone when they’re provoked so hard they can’t control themselves.
I think that even in the 50s, the majority of people would say that it is not morally right to coldly and deliberately stalk and kill the person your spouse is cheating with. Maybe I’m wrong, but I don’t think so. They probably wouldn’t say it’s as bad as murdering a nun or something, but that it’s still wrong. Most people certainly considered murder to be a sin.
Now, you can of course argue that people don’t really go “temporarily insane” or literally lose control of themselves. I don’t know.
There are also ways I can see this being applied that would probably not be endorsed by the general public yet would logically make sense. For instance, suppose that a conservative Muslim father walks in and finds his daughter in bed with another woman, a Western woman. Maybe he goes “temporarily insane” and kills his daughter in the heat of the moment, since he’s so full of anger and shame.
If he’s really “temporarily insane”, then that act as such wasn’t his fault. His culpability depends, I suppose, on how much you think he is to blame for being the type of conservative Muslim who would have such a reaction. (You could say the same about the frame of mind of the Western man who murder’s his wife’s lover.)
Maybe each one of these killers is a public menace and needs to be locked up or put down like a mad dog, but that’s different. You don’t think a mad dog is morally blameworthy.
At common law the analogous rule was sudden provocation and only mitigated murder to manslaughter.
I doubt many people would verbally endorse stalking and killing one’s spouse’s lover, but how many would vote not guilty? I don’t know about the 50s, but a lot of people do that in Texas today.
@ Douglas Knight:
Do you have sources to back that up or just stereotypes about Texas frontier justice?
My impression is that juries are typically far too quick to convict people who really shouldn’t be convicted, and that nullification is quite rare. This is partly because the government removes anyone from the jury who doesn’t swear that they would convict according to the letter of the law.
Of course people can lie about this (and Michael Huemer argues that they are morally obligated to lie and nullify in e.g. drug cases), but many people feel they shouldn’t.
Well, one could certainly argue that putting a thumb on the scale is not a proper function of government — the libertarian position, which most people regard as conservative. Or one could argue that there can be no such thing as a disinterested party to do the redistributing, that it will inevitably be “captured” by various interest groups. Or — and this is closest to the classical conservative ideal, I think — one could say that society as a whole is best served by allowing each person to achieve their own individual potential, and any intervention will result in a net loss of utility.
Looks are largely genetic. Murray would naturally not recommend better looking people be disfigured or to somehow re-distribute their looks (assuming technology for this existed – face transplant?) but he will likely recommend not very good looking people to watch their diet, dress better, and for god’s sake use some make up.
Similar with health …..
This is a nitpick.
Isn’t this kind of exactly backwards? The genome encodes extremely low-level processes (protein formation), which give rise to cells, which after several layers of abstraction give rise to the high-level behaviors we call “intelligence” or “conscientiousness”. The genome doesn’t encode the latter. Strictly speaking, it doesn’t even encode individual cells, but only the proteins which make up those cells.
I think there are elements of both. For example, my developmental biology is only at first-year[1] undergrad level, but IIRC there are processes that occur fairly early on in development that create gradients that sort of sketch out what the whole organism is going to be like on a large scale, then individual cells use that grid information and signals from neighbouring cells etc. to decide how to develop.
There’s a mess of different things that could be considered high-level or low-level, and that gets tangled up in such other things as global/local. If we think about signals, whether they’re calcium ions, sex hormones, transcription factors or whatever[2], some are at the head of long branched-and-rebranched causal chains (causal trees!) which have far-reaching[3] consequences, some are much closer to the leaves of the causal trees and have quite narrow, specific consequences. So there’s upstream/downstream too, and it’s easy to think of upstream things as being both prior to (by definition) and higher level than downstream things.
[1] Of course, what “first-year level” means is quite variable; I hear that at some universities, the first year is basically a revision of A-Level. But I digress.
[2] To be pedantic: variations in the levels/distribution/etc. of said entities. A signal could be thought of as a difference that makes a difference.
[3] But not necessarily all of the consequences all of the time, if some of the proteins etc. that handle the things downstream have unusual variants. One of my favourite examples is Complete Androgen Insensitivity Syndrome, where the signals originating from SRY1 (the main gene on the Y chromosome) get as far as stimulating testosterone etc. production in the gonads but the signals from that go unrecognised due to non-functioning androgen receptors, usually with the result of a female phenotype. That said, the androgen receptors affected in CAIS may be downstream of SRY1 but they’re pretty far upstream in the big picture.
I’m just happy I was able to express this in a way different enough from the Zakharov quote that nobody noticed I was plagiarizing it.
I noticed. We must dissent…
No, we must consent. But only if our score is high enough.
Damnit. Now I want to play AC again.
Resources exist to be consumed. And consumed they will be, if not by this generation then by some future. By what right does this forgotten future seek to deny us our birthright? None I say! Let us take what is ours, chew and eat our fill.
***
I plan to live forever, of course, but barring that I’d settle for a couple thousand years. Even five hundred would be pretty nice.
NOT HELPING. 🙁
Depends on whether you look at it from an epistemic or ontological perspective.
Is quantum physics high-level or low-level. It’s high-level in the epistemic sense that it uses highly-abstract math and is far-removed from everyday experience (have you ever actually seen a proton?). It’s low-level in the ontological sense that it subvenes reality.
In asking whether cosmology is high-level or low level, we don’t get an impedance mismatch because it’s high-level in both senses of the term. But that’s just coincidence.
I’ve always understood “high level” to mean exactly what Scott used it as – “big picture” stuff that doesn’t get into the weeds, so to speak. Never heard it used in any other sense – when one speaks of high-level strategy, you’re not talking about how individual fire teams are moving to the objective; when you speak of high-level programming, you’re talking about abstract-level stuff, that frequently looks like natural language.
I’ve always interpreted non-shared environment as basically developmental randomness that’s very hard to pin down, and I would have expected school to show up under shared environment anyway.
From what I remember of The Nurture Assumption, Judith Harris makes a case that peers are an important component of non-shared environment, but doesn’t present very strong evidence for that (compared to her evidence that shared environment, and thus parenting style and most common interventions, doesn’t play a big role). It’s more like “there’s this big glob of unidentified factors, maybe peers play a part?” (with a bit of evidence, but nothing super strong).
Apart from Haris, I haven’t heard anybody attribute non-shared environment to peers (and even she doesn’t say that it’s the biggest part of that), and I don’t remember anybody assigning it to school.
I remember thinking at the time that if you managed to find the same sort of peer group regardless of who your parents were — probably the peer group was the consequence, not the cause.
What is funny about some of these discussions is that while it is the left that is somewhat against genetic “determinism”, one could characterize much of the Left-wing program as “increasing heritability.” I would doubt Bernie Sanders would come out and say it this way, but it seems clear that what he is after is a society where income is .90 heritable. Or, to put it in another way, if randomly swapping your parents, schools, neighborhood, (even, at the margins, your race) doesn’t make much of a difference, then this is a good society from Bernie’s POV.
Many libertarians might agree that this is their ideal too, they just disagree on the means to get there (and some, but not all, left-wingers may add a dose of egalitarianism into the mix).
Not sure how you reach that conclusion about Sanders. Some in the left want equal opportunities (like tuitions paid from taxes which arguably could increase heritability-income correlation), but a bunch of others argue for equal outcomes (they want things like minimum wages, tax-payer health care, progressive taxes, etc) which decrease heritability-income correlation.
I didn’t mean to single out Sanders specifically, but, yes, tuition free college is exactly the kind of policy that would increase heritability. Free healthcare for children certainly increases it as well, as does free childcare. Cheaper housing would mean that where you live is less determined by your “ability” to inherit. Better worker insurance means less variation due to the random noise of accidents.
Progressive taxes may decrease variability in income, but I’m not sure this would decrease heritability. More equal outcomes may still result in a society with high heritability coefficients. You are decreasing variability on both parts of the ratio, but I don’t see that the ratio changes in a consistent way. I see it as an almost orthogonal effect.
By the time they get to college age, their environment has had a long time to work on them:
http://www.city-journal.org/html/lost-ghetto-12261.html
Wait, isn’t this backwards? If changing who my parents are doesn’t change my income, then I didn’t get my income from my parents, right?
Are we talking biological parents or legal parents?
I thought biological, but does it change the answer either way?
I rather think the OP meant legal.
Assumption: Income is genetic.
If you change your biological parents, then you’re a different person, and your results are consistent with your new ancestry. If this doesn’t change your income, then your new ancestry is the same quality as the previous.
If you change your legal parents, you still have the same biological parents, so it changes nothing.
This of course comes from “heritability” having a counterintuitive defintion – it counts variation relative to some background and if the background changes then so does the heritability.
Suppose some allele somewhere has a “cash value” of $1000 of lifetime earnings compared to some other allele. If we reduce environment-caused variation (in such a way as to decrease financial inequality), then the cash value may stay the same – however the decrease in financial inequality means that the relative contribution of the gene to financial inequality goes up even as its absolute contribution stays the same.
I identify as centre-left (and those policy-matching quiz things say I’d be a Bernie supporter if I were American – then again, if I were American I might have different views – I’m a liberal/social democrat, both literally and metaphorically a Lib Dem); these issues bring out my lefty side. “Increasing heritability” is not what really motivates; if someone increases the genetic-caused contributions to income inequality while keeping enviromental contributions the same, that’s not a win, far from it. The things that might be wins might also have the side effect of increasing this strange thing called “heritability” but that’s not why we want them.
>The things that might be wins might also have the side effect of increasing this strange thing called “heritability” but that’s not why we want them.
I think that’s the crux of the issue. In pursuing and achieving their goals, the left incidentally increases the relative importance of inherited traits, as a side-effect.
A lot of the ” genetic determinists” of the 20th century had progressive or socialist political views, actually. So Bernie’s POV for a good society isn’t a new one for leftists. The difference is that many of the earlier leftists endorsed forced/coercive eugenics in order to form this “good society”.
Modern leftists instead take a Rawlsian view that people who have the abilities to succeed can have the chances to do so, and those that don’t for whatever reason are aided. Hence, support for better access to higher education to make sure kids w/ the ability aren’t getting left behind http://www.nber.org/papers/w17633.pdf , and also for a welfare state
Rawlsian: Well, in my case, sort of. I count myself as Not A Rawlsian for various reasons (for example I don’t like his stipulation that you’re not allowed to know or act on probabilities and “therefore” must use maximin, I’m more happy with the Harsanyi version of the veil that allows people to go for maximum expected utility and therefore works out as an interesting form of rule utilitarianism, also various other things about A Theory Of Justice wound me up) but the basic veil idea appeals to me.
Then again, I’m odd. I grew up as an 140 IQ Aspie (without diagnosis until a few years ago) who aced maths homework in double-quick time and always got picked last for sports (and picked on by bullies). My weird mix of abilities and disabilities really felt – and feel – a lot like luck of the draw. With the IQ thing… on the one hand I’m almost embarrased by how much cash I earn[1] and am one of those annoying people who think they don’t pay enough tax, OTOH some lefty types (more the SJ than social democrat side of things) who go into “well it’s all about which school you went to and how rich and middle-class your parents are” really irritate me.
Looking at my basic feelings, I feel that inspiration and talent should be rewarded with recognition and interesting and exciting challenges, and perspiration should be rewarded with cold hard cash. I’m not sure how this fits into philosophical theories or possible economic practises.
[1] Except when I think “I’d like to buy a flat in Cambridge” and then I think maybe I’d like some more money. But if everyone else in my position had more money too, flats would be more expensive and I’d get no benefit.
The minimum wage case is tricky. People who argue for a higher minimum wage believe that it would reduce variance in outcome. But if the actual effect is to price unskilled workers out of the market and make them permanently unemployed it would have the opposite effect.
I agree that Harsanyi makes more sense than Rawls, but maximizing average utility has implications that I doubt you would accept.
Consider Mead’s example. There is a world with two cities, A and B, that have little interaction. People in both cities are happy, but people in A are even happier than people in B. If everyone in B is painlessly killed, average utility goes up.
For one of my old examples, before I discovered Mead’s, imagine that someone is stranded in youth on an island and lives his whole life there without ever interacting with another human being. Does it make sense to say that whether his existence is a good thing or a bad thing depends on how happy other people are?
Of course this depends on what you mean by average utility. If you mean lifetime average utility – i.e. total utility over all time divided by total lives lived over all time, then this doesn’t apply, if all of the lives in B are cut short there’s a loss of from the numerator with no loss from the denominator. I think that framing is in keeping with the Rawls/Harsanyi thought experiment – if you’re some disembodied spirit knowing you’re going to be born into the world but not where or when, you want to avoid having your life cut short.
Also, having a contract hanging over you where your life might get terminated for not being happy enough is not likely to be pleasant so a contract where that was a danger might not maximise utility even if you went for a different definition of average utility. These sorts of effects are the reason I call it an “interesting” form of rule utilitarianism. The differences between a contract that maximises utility and a contract that says “maximise utility”. Especially, you really don’t want one that says, “maximise utility, or forever be tormented by your conscience. Oh, BTW, this contract doesn’t magically give you the strength of character to actually maximise utility. Good luck!”
Of course if you don’t know you’re going to be born into the world, if being born into the world is potentially a rare and special privilege (provided that life is worth living), then you’re looking at something more like total utilitarianism. Of course, there’s an odd thing relative to my variant of average utilitarianism. Suppose someone is living a kind-of OK life which could be better, and there are limited resources which means the population is limited. Under total utilitarianism there’s a temptation to “chop and change”, to kill the OK person and replace them with a happier one. Under my framing of average, there’s no sense in cutting short an even marginally worthwhile life (unless externalities put their thumb on the scales).
Bottom line: I’m sort of vaguely utilitarian-leaning, or sort-of-drifting-Kantwards-from-a-vaguely-utilitarian-start-point, but I don’t have firm commitments, also utilitarianism has a lot of variants and it’s interesting to consider more than one of them – they seem to be doing different things, mapping to different sorts of moral feelings if that makes sense. I don’t expect foundational issues to be sorted out conclusively in my lifetime, and in practise I think what’s more important is what you build on top of the foundations, and what effects that has. But I would say that, I’m utilitarian-leaning!
“Those with ability are not left behind” is exactly the motivation for policies that increases heritability.
yes I’m agreeing with you!
The Eugenists were progressive and humane souls who deplored the willingness of conservatives to let people with bad genes continue to propagate and degrade. Obviously the only way to go was involuntary eugenic sterilization.
“Three generations of imbeciles are enough.”
Were they wrong in principle or only in execution?
They certainly tried to make the execution as modern, quick, and efficient as scientific and technological progress would let them.
The problem I often hear about the early eugenics programs was that they started targeting all sorts of people who weren’t congenital misfits (gypsies, blacks, religious minorities etc.). The rest is all just guilt by affiliation with the nazis.
So, the question presents itself, is forced sterilization something that should never, ever be used, or just something that needs a more robust system of due process?
I tend toward the latter opinion. For instance, anyone too mentally challenged to be able to consent to sex under any circumstances is probably too mentally challenged to be having kids. Of course, they are also too mentally challenged to assist in their own defense, so I think you’d need a permanent legal structure to resist demands for sterilization. I think if it can be shown in a court of law that a person is A: incapable of caring for children, and B: at significant risk of having children, so long as there is a reasonable amount of due process, there’s little downside to sterilization regimes.
On the question of sterilization, I’m leaning towards the hard-line “no, fuck you” for the same reasons as the NRA does it, but I’m not altogether dissuaded from potentially extending the option of sterilization as an alternative to execution for those sentenced to death.
I’d be more open to the second possibility if the early 20th century Eugenics movement was the only time that something like this got out of hand. It wasn’t.
The usual alternative to a death sentence is life imprisonment without parole. It is already extremely difficult (though not impossible) for someone to reproduce under those circumstances, so sterilization would be redundant.
Yes, it would. I mentioned that option to illustrate my stance on this – that nothing short of deeds meriting the death penalty should be considered when deciding whether to sterilize someone.
Since the principle was “We get to breed people like sheep, for our interests not the sheep’s,” — are you prepared to defend that as “not wrong’?
Yes, but this demonstrates a common confusion about heritability. Heritability is the proportion of variance due to genes. If you eliminate environmental variation, ideally by making it all optimal, then outcome variance goes down. The remaining variance is due to genes and noise and if you emphasize that, it might sound bad, but that’s because you’re focusing on the dimensionless heritability rather than the substantial variance.
@Douglas Knight. “If you eliminate environmental variation, ideally by making it all optimal, then outcome variance goes down.”
This is not necessarily the case. Depends on whether the variance of the genes was larger or not than the variance in the starting (pre-optimal) environment.
More noise may obscure and mitigate the signal strength/variance, less noise may make it more pronounced. Varied environment introduces a lot of noise (which may or may not be distributed in such a way as to exaggerate signal).
Suppose people are inherently very different with respect to their genetic propensity to run fast. Suppose further that in a pre-optimized environment many runners are malnourished, or really busy, or culturally compelled to not run (running is for losers…), or there were castes of people specializing in running (regardless of genetic endowment). Or maybe only rich people (only a fraction of whom are genetic good runners) are allowed/able to practice running.
Optimizing this environment, whereby everyone is encouraged and trained to run as fast as they are genetically able to, will yield larger outcome differences in speed than the pre-optimized environment.
No, it does not depend on that.
It is theoretically possible that optimizing the environment can increase variance. However, it is almost never true and certainly not in your example of running. It is only possible if the new, optimal environment had low measure in the old distribution.
To clarify, when I wrote “swapping your parents” I had in mind the unethical experiment of swapping babies in a maternity unbeknown to the parents. In a society such as traditional India, the kid’s futures would be determined mostly by who their legal/social parents were. In Scandinavia, mostly by who their biological parents were.
It’s not just about income either. In our current societies, it is still very unequal whether it is safe (for varying degrees of safe) to come out as not having a standard sexuality/gender identity. Naturally, this leads some people to hide/repress their inner identities.
BMI should also be more heritable. Wouldn’t a better society have had better school meals for everyone so that all kids would have tasty and healthy food?
One more clarification : I didn’t mean that left wingers are motivated by increasing heritability.
Nobody is motivated by decreasing the Gini index either (how many people can remember or even recognise its definition?). Instead, one finds it morally reprehensible that some people have so much more than others and woshes to decrease this gap. The Gini index just happens to be a pretty decent single number measure of this fuzzy concept.
Others find a moral value in ensuring that everyone has the same opportunities in life. Heritability just happens to be a pretty decent single number measure of this fuzzy concept.
So, you’re saying that leftists can/should use heritability to measure the progress of their cause?
>In a society such as traditional India, the kid’s futures would be determined mostly by who their legal/social parents were. In Scandinavia, mostly by who their biological parents were.
Have you read “The Son Also Rises” by Gregory Clark? If not, I encourage it. His thesis is that if you do that – move kids randomly about, even in a stratified society as India – the cream will on average rise to the top anyway.
The left-wing agenda has both heritability-increasing and heritability-decreasing elements. For example, universal access to high-quality schools and health care would tend to increase heritability, but progressive taxation, affirmative action, labor unions, and large public sectors would tend to make genetic differences less important.
@JK: “…..but progressive taxation, affirmative action, labor unions, and large public sectors would tend to make genetic differences less important.”
True in theory but rarely true in practice. As they say, some people are more equal than others.
I grew up in a communist country and even though everyone was in a union and there was no private property and school and healthcare was free, people were still hierarchically ordered by (largely genetic) ability. Access, position, and power simply played a larger role than pure money. Sometimes a lot larger – think being able to do things that even a lot of money would not let you do in a modern democratic, capitalistic society.
studies of many different traits tend to find that ~50% of the variation is heritable and ~50% is due to non-shared environment, with the contribution of shared environment usually lower and often negligible
I would find it extremely unlikely that somehow shared environment is negligible, and yet non shared Was Very important. How would it be that different parents make little impact but different schoolmates have a large one? A very good explanation would be needed.
A simpler explanation is what you point out: they are not measuring different parts of the same thing (environment) but rather other stuff.
Yeah
>How would it be that different parents make little impact but different schoolmates have a large one?
More likely that neither make a very big difference. NSE is a huge bucket, and it’s not necessary that the factors in it that humans intuitively think are important actually are important.
The idea is that people try to emulate their peers more than their parents. It’s why the children of immigrants dont have their parents accent.
The explanation Judith Harris offers is that humans are good at realizing that different environments have different rules. The child growing up has one personality at home, a different one interacting with his peers at school. It’s the latter that develops into the adult personality, since as an adult one is interacting mostly with peers not parents.
Or perhaps the personality attracted this particular peer group.
Regarding #1: don’t studies already correct for this when it comes to things like IQ? We know how IQ scores distribute around the real value, right? So it should be trivial to work backwards and correct the heritability number for the error.
You mean do https://en.wikipedia.org/wiki/Correction_for_attenuation ? Yes, it’s a good idea, but by definition you don’t know the exact correction in the twin population in question (and you can see even on this page that a favorite reaction is to question the representativeness or external validity of estimates on a twin population) since if you had the multiple measures which could give you the correlation between the latent variable and measurements you would just use them directly; and it’s one of the extra refinements that authors typically don’t get around to even if it would improve estimates some. (Time and energy is always limited.)
They also don’t correct for assortative mating (AM), which deflates the A estimate for traits (AM makes DZ more genetically similar, but not MZs who are already maximally similar).
Not correcting results is fine, they are just the uncorrected values. A meta-analysis could correct the values (cf. Hunter and Schmidt’s book). The problem is when they start theorizing from the values derived from data uncorrected for measurement error and AM, which are heavily in favor of finding lower A and more C/E. When measurement error is high, this usually results in theorizing about the importance of non-shared environment.
Meh.
>but by definition you don’t know the exact correction in the twin population in question
I don’t see why not. If a person’s “real” IQ is x, the test scores will be ~N(x, s) right? Can’t be that hard to figure out s, then you can calculate the “separation index”. Am I missing something?
Because you’re estimating the reliability based on a different population, and as you can see on this page, people often believe that twin populations differ from regular populations. To estimate the reliability based on the twin sample, you would need multiple measures, and in that case, why are you bothering to correct one test measurement when you can use a SEM on your multiple measures to estimate the latent variable more accurately? eg if you do 5 different IQ tests, sure, you can estimate the reliability of any 1 of them and correct for disattenuation, but why do that instead of loading the 5 scores onto a latent variable estimate? So you either need to make an assumption about reliability transferring which people can attack you for, or you are throwing out data.
> why are you bothering to correct one test measurement when you can use a SEM on your multiple measures
Because to correct one test measurement is way cheaper/easier/faster. You only need to give the additional tests to a relatively small subsample to get a good estimate of the variability (in fact someone has probably done this already and you can use their number). The alternative requires you to give multiple tests to everyone.
>So you either need to make an assumption about reliability transferring which people can attack you for
Doesn’t strike me as an outlandish assumption.
I am not sure the distinction between error and luck is neatly kept. The example for error is a person making a lucky choice on a multiple question exam. Wouldn’t error work better on things like the lab assistant being tired and grading the same exam differently?
I’ve been noticing for a long time that they declare “environment!” whenever they don’t know what causes it.
I don’t see how “2. Luck of the draw” is not an example of non-shared environment? Sure, it is not about schooling and similar, but how else would you categorize twins ending up in different work environment?
It is non-shared environment, I just want to make it very clear that it’s not about non-shared environment permanently affecting somebody’s disposition/personality/brain/whatever.
Why having a jerk boss would affect you any less permanently than having a bad teacher in principle (rather than in degree, due to you being of older age or length of exposure to the boss/teacher)? But I think I see what you are saying – you want to make sure people understand what might get lumped into this big category of “non-shared environment”. But frankly, at least speaking for myself, I would have assumed this and other example were included in the definition to begin with.
A jerk boss would have a direct boss -> income effect, in addition to the boss -> personality -> income effect that would classically count as “non-shared environment”.
Scott, your reasoning is sound and it’s extremely obvious why experts in these fields don’t want to talk about it. They evade this reality for the same reason that ridiculous “blank slate” and “social construction” theories of cognition persist despite all the evidence against them.
The largest single blocker is that really facing up to what the data tell us about human development would force a lot of people to abandon their emotional commitment to the idea that we can social-engineer our way to a better world through wise and targeted environmental interventions. And, they’d have give up on all the opportunities for mutual congratulation and virtue-signaling that belief entails. The horror! The horror!
Beyond that, I think there’s an even deeper fear that if the replicable factors in human variation are dominated by genetics, the racists and eugenicists will turn out to have been right after all. That possibility is so terrifying and disgusting that it can fuel arbitrarily large amounts of willful self-deception.
This theory leads to a testable prediction: when we get the ability to patch in germ-line improvements in traits like IQ, time preference, and impulse control, ideological resistence to hereditarianism will abruptly collapse and “non-shared environment” will be correctly interpreted.
I don’t think your single largest blocker matters much. It’s all the second thing.
It was a pretty bitter disappointment when I realized just how many of my fellow leftists were being held back from being undesirable purging fascists, not by fundamental disagreement, but rather by a mass game of pretend.
This should all be considered subtext of most discussions on genetics. I don’t think it’s too productive to actually come out and say it.
(also, most of the social engineering falls into shared environment and so isn’t super related to this topic)
I think it’s generally helpful to point out when people have biases and why. If it’s not, then the entire edifice of Yudkowskian rationalism is erected on a false premise. And if what you mean by “not productive” is “it makes people uncomfortable”, you should be nearly the last person I have to remind that biases like that are particularly important to analyze and root out.
A hostile commenter one accused you of being “very protective of his liberal homies”. Please don’t disappoint me by actually living down to that criticism, Scott.
And oh, no, nobody gets to evade the problem by pointing out that most social engineering falls under shared environment. The crux of the problem is that under an unblinkered interpretation of “non-shared environment”, the joint effects of genetics and non-shared environment are so dominant that it’s hard to see how the effectiveness of social engineering can be anything but a delusion in the minds of would-be engineers.
Of course, to the people Thomas Sowell refers to as “the Anointed” or “Teflon prophets”, this is unthinkable. So entire scientific disciplines have to be distorted out of shape to preserve their belief that their redemptive mission is possible.
I may be missing something important here, but if we measure the impact of something on a variation of outcome, we will not find anything if there is no variation among the inputs. And “social engineering” usually tries to apply to everyone, which is how (if done successfully) it would avoid measurement.
To illustrate, here is a toy model: suppose that human IQ is completely determined by two factors — the smart gene, and the consumation of lead — according to the following table:
smart gene + no lead = 130 IQ
no smart gene + no lead = 100 IQ
smart gene + lead = 70 IQ
no smart gene + lead = 40 IQ
However, a successful social engineering campaign made lead illegal and completely removed its presence from the whole country. There is not a single piece of lead to find anywhere. So currently the whole population is:
10% with smart gene and 130 IQ
90% without smart gene and 100 IQ
Now a group of scientists studies this population, and comes to conclusion that all differences between these two groups are caused by the smart gene, i.e. it is 100% nature and 0% environment.
(Ironically, in a parallel universe, where Adolf Einstein and his Mensastaffel commandos sent everyone without the smart gene to concentration camps, but the lead in environment was mostly ignored, the scientists currently find out that intelligence is 0% nature and 100% environment.)
Yes, exactly.
I’m going to partly steelman your argument by pointing out an obvious historical case of much better outcomes by changing the environment for everybody: the near-elimination of epidemics in the First World by public sanitation measures.
This, however, is not what is usually meant by “social engineering”, and the divergence is why I snarked about targeted interventions. Outlawing lead, or routinely laying sewers and plumbing for clean water, are not targeted interventions.
What’s being protected by the misinterpretation of “non-shared environment” that Scott pointed out is the notion that we can target a disadvantaged group, wisely design a measure to rectify the disadvantage, and then expect that group to achieve some measure of parity that we specify without running into any innate disadvantages that torpedo the program.
Examples of how this sort of program fails are legion, but hotly denied. When Larry Summers pointed out that difference between male and female dispersion of IQ predicts fewer women in the region of the right-hand tail that produces star STEM talent, he was mobbed. Nobody wanted to talk about the facts, just pound on their insistence that we have to intervene harder and more specifically.
Another relevant example is targeted anti-poverty programs versus measures that reduce the overhead of employment. This experiment has been run many times. When the overhead of employment goes down, the amount of value employers and employees can capture from the exchange goes up. Employers create more jobs and people self-selected for the capacity to do it work their way out of poverty.
On the other hand if you try to target an entire disadvantaged population and lift it out of poverty, you are very likely to find out that there are intractable reasons that population is poor, reasons that aren’t part of the shared environment your social engineers can manipulate. Low average IQ, high time preference, poor impulse control, and genetic susceptibility to alcoholism are hardy perennials here (and I’m thinking of my 19th-century Irish ancestors as much as any hot-button contemporary examples).
There’s a sting in the tail. The effect of successful targeted “uplift” is both dysgenic and dyscultural. We harvest the bright kids out of poor neighborhoods with scholarships. Does anyone consider what that does to the breeding population and the culture they left behind?
Sewers constitute an intervention that was targeted at the health-disadvantaged group: the urban. And it did compress variance, reducing C and increasing A.
Similarly, the elimination of various parasites, like hookworm, was quite consciously targeted at the disadvantaged: the South. It compressed variance (of income, and, I think IQ), reducing C and increasing A.
Though one thing that nobody ever brings up is that there are countries where the STEM gender gap, at least in education, is lower, non-existent, or even reversed. However, the reason why no one ever brings these countries up becomes clear once you look at which countries these are.
For example, 42% of Indian Computer Science majors are female. India is, of course, a far more sexist society than the US, what with the skewed sex ratio (10 boys for every 9 girls under the age of 6) due to rampant sex-selective abortion and infanticide, the thousands of dowry murders every year, et cetera. An even more extreme example is that 70% of all science majors in the Islamic Republic of Iran are female. Another society that was significantly more sexist than the modern US but had a smaller STEM gender gap was the US 30 years ago: in 1984 37% of Computer Science majors in America were female.
So there is evidence that social environment can impact how many women enter STEM fields. But the same evidence does not support the idea that simply “eliminating sexism” would “fix” the STEM gender gap.
This is the sort of thing I think benefits from noting that “right” is insufficiently defined. It’s entirely possible for the racists and eugenicists to be factually correct about IQ variation et al (“e.g. ‘the average IQ of population X is lower than the average IQ of population Y”), wrong about the moral consequences of those facts (“therefore group X has less moral worth than group Y”), and vile with regard to what should be done about it (exterminate! exterminate!).
“Fear that the hated Other Side will turn out to be right” conflates these three meanings.
Except that where the racial issue actually comes up, at least in the society I live in, isn’t whether one group is less morally worthy than another. It’s whether observed differences in outcomes are due to discrimination. It’s treated as a matter of faith that they are, which depends on the (usually unstated) assumption that differences in innate abilities don’t exist or at least are insignificant.
It’s a particularly indefensible assumption in the case of m/f differences in outcome. We are “as if designed” for reproductive success, males and females differ precisely in their role in reproduction, so it’s unlikely that the optimal design is the same for both.
Which is why I argue that while some on the right say they don’t believe in evolution, many on the left demonstrate that they don’t believe in it by rejecting implications of evolution that clash with their ideology.
Though one should note that those aren’t the only two options. A third possible explanation is cultural differences. Recently I’ve started to lean towards the “culturalist” camp myself after reading about how successful recent African immigrants have been in the US and UK, including immigrants from highly unselected populations like Somali refugees. Both the hereditarian and discrimination* models have a hard time explaining that data, but it fits very well with the cultural model.
Not that it makes any real difference, of course, since a lot of people on the left consider discussions of cultural factors to be just as bigoted and evil as discussion of innate abilities. See their reaction to Thomas Sowell, Amy Chau, etc.
* Unless one were to absurdly argue that Somali immigrants who look black, are overwhelmingly Muslim, and have similar accents to the scary Somali pirates and terrorists that have been a major feature on the news for the past decade are not subject to discrimination.
The evolution of sex differences is a little complicated, however, by the fact that people inherit genes from their opposite sex parents. A trait that improves the reproductive fitness of one sex but is neutral to the reproductive fitness of the opposite sex could easily evolve to be expressed in both sexes, because there is no evolutionary benefit to restricting its expression to one sex. Which is not to say that no innate sex differences exist (clearly there is a very significant sex difference in average upper body strength), just that one should be careful when applying evolutionary models to this sort of thing.
But again, this issue is really hard to talk about because advocates for the discrimination model tend to conflate discrimination in adults with sex differences in childhood and adolescent socialization, even though they are very different phenomena.
Not to mention that “culturalism” is the mainstream conservative position on the issue of different outcomes among racial groups.
Of course, many on the left regard “culture” as a dogwhistle for “innate racial differences”. To some extent, it probably is. But for the most part, I think it’s the honest belief of most mainstream conservatives.
What’s a little bit funny is that the proponents of “discriminationism” seem to regard “geneticism” as the worse enemy. However, the genetic view seems more amenable to the left-wing position that the lesser success of some groups is not their fault.
“Culturalism”, in contrast, seems to suggest at least to some extent that individual members of disadvantaged minorities have it within their power to succeed if they “take initiative” and reject their cultural norms. So it’s literally a victim-blaming explanation. Not wholly, of course, because they’re not responsible for being born into their culture. But people have more ability to change their culture than their genes or the discrimination of society.
Your point about African immigrants was made long ago by Sowell in _Ethnic America_ in the context of West Indian immigrants.
As a long proponent of culturalism, I support those theses. Genetics matter to the individual, and perhaps to the tails of the bell curve.
Culture matters for the average, and the whole. Athenians were no different genetically than the Spartans, but the cultural differences were vast. And so were the results.
Come to think of it, the issue is still treated as a matter of moral worth; it’s just that the group which gets the good outcomes is treated as the morally suspect one.
Correct. I think that people who misinterpret “non-shared environment” as Scott describes suffer from exactly this conflation.
I am dating a twin right now: She’s really into kink whereas her twin sister, not so much.
Also the twin works for $largeTechCompany whereas the one I’m seeing works in fashion.
Not sure where I’m going with this, but I can say that I was a bit surprised when she told me this; thinking back to all of those twin studies that say that twins who are separated at birth usually end up with the same sort of jobs/income levels/preferences.
I think siblings begin to (weirdly) grow more similar as they separate and grow older. When my brother and I left home my mother would notice that we would regularly have almost identical conversations with her. I was always the nerdier of the two and he was the jockier when we were in the same house, but as soon as we were separate he basically lost interest in sports.
I think this points to genetic “destiny” becoming more pronounced the older one gets, which meshes with the finding that many early interventions produce only temporary improvements. Think about how babies look much more similar to one another than old people, whose features tend to become more and more pronounced.
I think of it almost like: if you take a ball with a characteristic set of irregular bumps and pits and throw it down a hill, then those irregularities will tend to, in the long run, determine the direction it goes and where it finally ends up. If you throw it in a different direction or put a spin on it or it runs into some bumps on the hill, all of those things can alter the trajectory, but they don’t alter the built-in bias the object has to roll in a certain way. And the longer the roll, the more the shape will determine the position and trajectory and the less the initial throw and bumps in the road will matter.
And, of course, you can guess based on the shape of the ball where it will likely break, though as to which bump in the road causes the break and how long it will roll before breaking, one is less certain.
Makes some sense; if you are quite similar by nature, but still want to differentiate yourself in the family unit you may well emphasize different interests or affectations, but when apart and finding a place in the wider world you won’t have the same pressure to go against your quite similar natures.
Although me and my closest sibling have always been very different.
Now I’m imagining educational porn about this debate.
Some of that, I think, is down to the desire of twins to differentiate themselves. Being treated as the same person times two (the dressing alike, giving them the same hairstyles, etc.) is cute for a while but as they get older, it’s tiresome.
So identical twins growing up together have a great motivation to emphasise the differences between themselves, to the point of completely different interests, jobs, and so on. Sue likes ballet so Ann likes mountaineering.
Whereas identical twins separated at birth or somewhat later don’t have a ‘mirror image’ they need to distinguish themselves from, and thus if Sue likes ballet, Ann might like ballet too.
Identical or fraternal twin?
This post is pretty good! You hit all the points properly and well. I’ll have a more detailed reply later.
Steve Sailer, at some point, came across an article which lambasted twin studies because there were a bunch of sources of error, similar to the ones you’ve discussed. (One thing you can add to the generator error, for example, is that people mutant while growing from a zygote to a full person, and this will happen differently for the two twins.)
But at the end of the day, Sailer observed, error means that the twin studies underestimate the effect of genes, whcih was the opposite conclusion from the one the author wanted to draw.
For 3d you meant zygote, not egg, right? There are eggs that split and get fertilized by different sperm (for 75%relatedness), but they’re not common.
Twins compete for resources in the womb and studies that don’t take this into account will overestimate the importance of the post-birth environment.
Very little. http://link.springer.com/article/10.1007/s10519-015-9745-3
I’m confused why a distinction is made between “shared environment” and “non-shared environment”. What would it mean to say that 25% of variability is due to shared factors and 25% due to unshared factors? ie how can a shared factor cause variability?
Isn’t this akin to dividing the effect of genetics into “shared genes” and “non-shared genes”?
Compare twins that grew up together to twins that grew up apart. The difference is the shared environment effect.
Thanks but I still don’t get it. Why don’t we just call that ‘environment’? How can a shared factor explain variance? Apologies for being dense.
To be pedantic, it’s variation in shared factors causing variation in outcomes. And the variance is the variance within the population at large, not the variance within the family. Some of those factors tend to be shared between siblings, either by genetics or shared upbringing, and some of those tend not to be.
So, one of the factors affecting, say, my income was how strict my parents were. People have parents of varying strictness, so there’s variance explained by the strictness factor. But that factor also applies to my sister. So if you knew my income, you could make a better guess at my sister’s income than if you didn’t, but not a perfect guess, because there are some genes and environmental factors that we don’t share. If you plot a scatterplot of person’s income vs sibling’s income, you get an elliptical splodge on a diagonal, and the correlation coefficient R^2 is the proportion of the variance[1] explained by shared factors.
[1] Note that variance is a bit of an odd quantity because it has odd units – variance in income is measured in square pounds (etc.) which doesn’t seem to correspond to anything. The plain English word “variation” is much better captured by standard deviation, which is the square root of the variance.
thanks
Shared environment causes differences between families. Unshared environment causes differences within families.
http://amzn.to/1PehrCL has some interesting points. Variation due to environment is only measurable to the extent that environment varies across the sample population. Height is ~90% heritable within a given population but South Korea has seen a 4 SD increase in height over the last few decades due to changes in environment. There is a sort of motte-and-bailey going on between ‘how much of the current variation is due to non-shared environment’ and ‘how much effect could we produce by altering the environment’.
Nisbett argues that adoption studies tend to underestimate the effects of environment on intelligence even within a given population, because adoption agencies filter for good home environments. He offers several studies which seek out unfiltered adoption and find much larger effects, and also a few early-life interventions which produce large effect sizes in life outcomes by improving the social environment in the first few years of life (eg http://onlinelibrary.wiley.com/doi/10.1111/j.1467-8624.1990.tb03559.x/abstract;jsessionid=644160C37704998A68314DF626DADD9C.f01t03).
I tend to discount Nisbett, since he not only denies, against mounting and overwhelming evidence, that heredity has anything whatsoever to do with the existing black-white IQ gap, but insists (bizarrely and contra factually) that IQ itself is highly malleable.
I dislike him viscerally as well, since when James D. Watson was driven from his post at the famous Cold Spring Harbor medical research laboratory for making politically incorrect remarks about IQ, Nisbett helped put the boot in, publishing an op-ed in the December 9, 2007 New York Times under the memorable title All Brains Are the Same Color.
Nisbett is a sell-out, though, unfortunately, a successful one.
Any other identical twins here? I only participated in one twin study growing up.
http://tennesseetwinstudy.org/styled/index.html
I don’t envy the researchers. I don’t know how they possibly obtain good information. I am sure this is my ignorance of the field. I only spoke with the researchers once. I only remember answering a couple questions for at most an hour. I was also in a fantastic mood that day since I was getting a small amount of money to spend on video games.
My twin and I are both very successful engineers. She owned her own house at 25. We both are in top 10 graduate programs in Computer Science. We still look very much alike, as when we meet strangers together they still ask us if we are twins. We are both very tall at 6’4. We are both bisexual. But in a lot of ways we are very different people. She’s super left wing where as I’m more center-right. I weight 50 pounds more than she does, since I lift weights and eat meat where she’s a vegan. She enjoys the city while I live in the suburbs.
But I also have the frustrating feeling that she is the only person in the world who actually thinks like me. Living in a city with more than its share of issues, we have mutual friends who are constantly making poor decisions or taking undo risks with there lives. We both can’t comprehend why our friends make these choices where to us they seem insane. I guess we are both super analytical and risk adverse.
I don’t know, being a twin is weird.
Not a twin, but the feeling that everyone else is a moron is pretty common. Even among morons.
I have it all the time. So many people incapable of seeing one move ahead on the game board, so to speak. I put it down to having a very high IQ, until I remember all the risky things I’ve done. I’m very good at avoiding some common pitfalls, but also at justifying those idiocies I happen to like.
How much of this is the fundamental attribution error? “Other people make stupid choices because they are stupid, duh. I also make a lot of mistakes, but that’s because I am tired, busy, or given wrong info; also some of my actions end up horribly because of random, difficult to predict, changes beyond my control.”
On the other hand, maybe everyone is quite moronic (compared with the hypothetical rational thinker), and the people with very high IQ are only slightly less so — and even they are usually under peer pressure or given misinformation by people with lower IQ than them.
Hypothesis: there are lots and lots of patterns in life. Everyone recognizes a small subset of them (with smarter people recognizing more). Then when we see other people not know the patterns we know, we think they’re dumb. They keep doing things that produce predictable negative results!
Of course, when bad things happen to us we don’t notice the pattern. So we blame them on bad luck. And other people watch us and wonder why we’re so dumb.
I agree with everyone here. I didn’t wanted to come off as a know it all. My twin and I aren’t better people because we make good money and live simple lifestyles well below our means.
My point was even if we have different views on life, it’s really special that I have someone to help center me when I do make mistakes. It’s also helpful that the person processes things like me, and really cares about me.
What was the timeline of your divergence? What came first among politics, urban/suburban, diet, and exercise?
Re. the microbiome thing: in addition to the overuse of antibiotics, which I think is a big problem (have any studies been done about rise of antibiotics and the increase in autoimmune disease?), I’ve heard that the increased number of C-sections may also be something of an issue. Unromantic as it sounds, the mother basically can’t help but poop on the baby a little when giving birth the regular way, whereas this infusion of mother’s gut flora is absent in the case of C-section.
Probably non-genetic biological effects play a role, but I think it’s unlikely that those more classic types of environmental forces are unimportant, because of two observations mentioned in this 2012 review paper on IQ:
1. The heritability of IQ has been found to vary significantly with socioeconomic status, with most research indicating the percentage of variance due to drops as you go down to lower levels of SES, as discussed on p. 132-134.
2. This observation on p. 136:
1)
Disconfirmed by meta-analysis for non-US studies and severely overestimated.
http://pss.sagepub.com/lookup/doi/10.1177/0956797615612727
2)
That’s only for children anyway, irrelevant for adult outcomes where C is ~0 for cognitive ability.
For 1), the meta-analysis you cite actually supports the idea that this effect holds in the US, it just concludes that the effect isn’t seen in Western Europe and Australia, perhaps due stronger welfare states (and I think most of these countries also don’t have local funding for schools like in the US, so the quality of one’s education doesn’t depend as much on the wealth of one’s local district). The abstract says “In U.S. studies, we found clear support for moderately sized Gene × SES effects. In studies from Western Europe and Australia, where social policies ensure more uniform access to high-quality education and health care, Gene × SES effects were zero or reversed.”
For 2), I’m not clear on why the value of c (shared environment) for adults should be relevant–isn’t this a question about non-shared environment, since they are comparing children adopted into high SES families vs. low SES families (which will naturally affect other things besides immediate family environment, like school system and neighborhood)? Maybe I’m looking at it wrong, but an explanation of your thinking here would be helpful. Also, it’s true most of these types of studies don’t look at IQ in adulthood, but this meta-analysis which was referenced in the review has a table of the different studies they analyzed on pages 304-306, it includes an “age at assessment” column which shows a large proportion of these studies do look at IQs of adolescents who were adopted as children, with a common age of assessment being 12-18 years old.
1)
And the study most people cite as evidence is a far outlier. Look at the dataset (which is public, read the paper), or here: https://twitter.com/KirkegaardEmil/status/676171601466040320
There’s also a new study for Australia with no effect.
https://www.researchgate.net/publication/295542367_When_does_socioeconomic_status_SES_moderate_the_heritability_of_IQ_No_evidence_for_gSES_interaction_for_IQ_in_a_representative_sample_of_1176_Australian_adolescent_twin_pairs
Meh. In general, interaction effect have a low prior and they are hard to find, statistically. In other words, there will be many false positives and much exaggeration.
2)
One cannot use studies of children that show C>0 to argue that environmental causes are important in explaining inequality among adults.
By the way, adoption IQ gains probably not what they seem.
https://www.researchgate.net/profile/Elijah_Armstrong/publication/266976086_Are_adoption_gains_on_the_g_factor_A_meta-analysis/links/561d251f08aec7945a252a55.pdf
Large Swedish transracial adoption study found that C has no effect on trans-racial adoptees’ IQ either.
http://emilkirkegaard.dk/en/?p=5663
If you want to argue that C is important for inequality, then cite this:
http://www.boydetective.net/wp-content/uploads/2014/07/BraniganMcCallumFreese_SF_2013.pdf
For educational attainment: overall A = 40%, C = 35%, E = 25%. C somewhat higher for females (Table 4, +9.6%points).
For 1), have you or anyone else tried re-analyzing the effect size if that one study is excluded? For 2), I hadn’t adequately understood the meaning of “shared environment”–googling a little about the concept I see that they don’t try to control for SES when measuring C, so that to say C=0 for a sample of adults implies that there’s no correlation between the SES of their parents and their adult IQ. But then the question I’d wonder about is, do the same studies that show C close to 0 for adults also specifically take note of what the SES of the participants was growing up, and make sure to include a broad cross-section that includes a significant number of adults who grew up fairly poor? The review paper by Nisbett et al. that I linked in my original comment argues on p. 136 that this isn’t usually done in the case of twin studies, “because lower SES individuals are difficult to recruit to laboratories and testing sites”, and that “even those that claim to have representative samples of the various social-class groups may have self-selection prob- lems: The lower SES individuals who volunteer may re- semble higher SES individuals on variables relevant to overestimation of heritability effects.” They also argue on the same page that the findings of C being close to 0 in adults aren’t all that consistent and that data on shared environment effects for adults are “sparse”:
I made a mistake above. The large Swedish transracial adoption study does not show that C=0 for the adoptees, it shows that parental education has no effect on the IQ scores. Still, it’s close to showing C=0 because parental education is highly correlated with parental IQ. However, it’s possible that some other parental variable has some effect. It would have to be something not strongly correlated with education otherwise it would show up.
1)
It won’t matter much, but feel free to DIY. The data are public as I said. So download the data, load them into R, load up the metafor package and analyze away.
2)
“I hadn’t adequately understood the meaning of “shared environment”–googling a little about the concept I see that they don’t try to control for SES when measuring C, so that to say C=0 for a sample of adults implies that there’s no correlation between the SES of their parents and their adult IQ.”
Sorry, but you still don’t understand it. There is a moderate to strong correlation between parental SES and offspring’s adult IQ scores, but this has nothing to do with estimating the C parameter in IQ scores (my mistake above of it seemed like this!). The values of C for IQ are independent of the correlation of parental SES x offspring adult IQ (for twin studies).
One can find studies that find non-zero C for adults for IQ. But there are also many studies with C=0. Due to measurement error and sampling error, if C=0, we do not expect to see C=0 in every study, but some variation around C = 0. It cannot be less than 0 when people estimate ACE using SEM, so the mean of C in studies will be somewhat larger than 0 even if it is all due to random variation.
See e.g. this recent review: https://www.researchgate.net/profile/Thomas_Bouchard2/publication/255692897_The_Wilson_Effect_The_Increase_in_Heritability_of_IQ_With_Age/links/545b97120cf2f1dbcbcafb26.pdf
Here’s a recent very large study.
http://www.nature.com/mp/journal/v15/n11/full/mp200955a.html
C ≈ .20 at age 17 and on a downwards trend since younger ages.
Possible the methodologically most sophisticated study is this:
http://www.sciencedirect.com.sci-hub.io/science/article/pii/S0160289614000099
Which found A = 86, E = 14, C = 0 for adults. N ≈ 1200 twins.
The adoption studies tend to find C = 0 for adults as well.
But, perhaps C is 10% for IQ at adulthood. My posterior is not 100% fixed at C = 0%. 🙂
–
Nisbett is an egalitarian, don’t take him too seriously. He is very biased in his citation of the literature in general. See e.g. his attempt at arguing for zero heritability of group (race) differences. It’s in the appendix of his most recent book (you can find it on libgen). If the goal is to counter-weigh the conservatives (like Lynn), reading Nisbett is a good idea.
If it doesn’t matter much, then were you not trying to discredit the conclusions of the original meta-analysis we were discussing when you pointed to that one study as an outlier? I thought maybe you were arguing that the meta-analysis included this particular study and that significantly affected its conclusions about the correlation between SES and IQ in the US.
Are you disagreeing with my statement that SES of the parents is included in the “shared environment” parameter C, or are you saying SES “has nothing to do with estimating the C parameter in IQ scores” for some other reason? If the former, I was basing my statement on this paper I came across, which says in the abstract that “When genetic similarity is controlled, siblings often appear no more alike than individuals selected at random from the population … it has become widely accepted that the source of this dissimilarity is a variance component called nonshared environment.” This seems to imply that for adopted children with different biological parents raised in the same home, they would be no more alike than any two random members of the population, which would mean the SES of their adoptive parents didn’t create any correlation. Furthermore, on the second page of the paper the authors write “Shared factors such as socioeconomic status, child rearing practices, and marital quality were assumed to affect siblings similarly and therefore to have little causal importance”, again implying socioeconomic status is supposed to be included in “share environment”. So if you do in fact dispute the idea that SES is considered to be part of “shared environment”, would you disagree with the authors or do you think I’m misunderstanding what they meant in these quotes?
Maybe non-shared environment should refer to things we can’t engineer genetically (e.g. whether you’ll work for a crappy CEO)
(1) as observed above, a well-designed study takes this random variation into account.
(2) You would expect things like mental disorders (which are extremely correlated with criminality) and tests like the SAT that measure a person’s basic cognitive function to be pretty nature-determined. But people who then turn around and say, “everything you care about is determined by genetics” drive me crazy. Let me try a clunky metaphor: say you’re an alien who can only see metal, and you want to study humans. You can only measure humans when they are driving cars.
So you do a bunch of studies and start getting excited: people have different styles of driving; after extensive research, you’ve learned to distinguish the unique driving patterns of a human even when he switches driving one car to another. But your subsequent results are kind of disappointing: all the things you can measure that you think are important don’t depend on the driver at all! The amount of mileage a driver gets mostly depends on the car. Determining max acceleration and breaking needed a well-designed study to factor out noise, but at the end of the day they only depend on the car. Traction mostly depends on the car. Maneuverability depends on the driver a bit, but when you get down to it, the minimum turning radius still depends on the car. So you conclude that everything really important about a human depends entirely on the car she is driving, and humanity is a boring race.
In psych studies, the things you can actually measure are, for obvious reasons, highly skewed towards measuring “motherboard-level” processing and obvious bugs. How many bits can a person remember? How many operations can they do per minute? Do they have a particular standard malfunction? And I think we are starting to accumulate evidence that these things are, to a significant extent, fixed: this is what they studies Scott linked seem to indicate. Now I’m not saying these measurements are unimportant. Many skills and careers require lower bounds on basic cognitive skills. Many malfunctions are debilitating and drive people to drug abuse or crime.
But in the few cases where you can actually measure something higher-level, I’d argue that you start seeing much more impact of “nurture”. For a more obvious example, take the impact of social stratification on practitioners of any creative profession (writer, musician, scientist): e.g., how many Jewish writers or scientists were active before 1900? (And note that Jews were already relatively rich by then, they just didn’t have access to what we consider a “liberal-arts” education — note that Einstein, as well as all the other famous early-20th-century Jewish physicists had the full liberal-arts education.) For a more nontrivial example, take different cultural approaches towards teaching math. There’s an often-observed preponderance of Iranian women in mathematics. There’s a lot of (well-deserved) buzz about the Russian math education system of the 80’s that produced an unreasonably large number of world-class mathematicians (http://russianmathblog.com/2010/10/what-makes-russian-mathematics-special/). None of these can be satisfactorily explained away by nurture, and these are just the examples I’m familiar with.
Basically, when people do heritability studies, it’s important to recognize that they are looking under the lamppost. This is a good thing (we can’t, at least at the moment, measure things like “kindness” or “good taste” without an unhealthy load of BS), but this streetlight effect is real and when people make sweeping generalizations about nature vs nurture without acknowledging this effect, it’s just .. wrong.
There was a pair of identical twin girls in my grade in middle school, K and J. K was very popular and well liked, and while J usually hung out with the same group of people as K by virtue of them being twins, she was often teased and excluded–I ran into her in the halls crying during recess several times. I wasn’t close friends with either of them, but as far as I could tell they seemed as similar as most identical twins are, except that J got shier and more introverted over the years because of the social environment. This is a kind of situation that strikes me as a “non-shared environment”, but then the question is, why was K well liked and J was shunned in the first place if they were so similar in almost every aspect? Probably just that middle school kids can be nasty for arbitrary reasons, but still, its interesting.
But there’s also a converse: not all nature represents what you think. If you live in a very sexist society where women are bared from employment, then there’s an almost perfect correlation between having a Y chromosome and a job.
One way to test this would be to distinguish between genes that cause, say visible racial traits, and those that don’t. If the genes for the first show up as having a high effect (relative to the second), then discrimination should be suspected. Versus the opposite conclusion if the more impactful genes don’t visibly or blatantly affect the outcome.
Another example: Robin Hanson keeps on pointing out the disproportionate advantage enjoyed by the good-looking. If a gene correlates strongly with outcomes and with beauty, but not with IQ, this would be imply that beauty discrimination is an important thing. And the impact would be “genetic”, but not in the way we usually think.
Twin studies don’t count race-related genes because all twins are the same race as their twin (except in extremely rare cases where an interracial couple has two dizygotic twins who look very racially different). The studies only count things which can conceivably differ between siblings. This would theoretically include X vs. Y, but I think they probably control for gender.
I’m reasonably sure that fraternal twins can appear to be of different races.
Yes, they can. Here’s some some examples.
I said that in a parenthetical right after my first sentence!
I’m sorry for the knee jerk reaction. Still, I wonder whether it’s extremely rare, or just somewhat rare.
Also, black people will get treated differently (by black people, not just by white people) on the basis of skin color, hair texture, etc., so having significantly different appearance could lead to different outcomes even if the difference is less than what’s needed for the twins to be considered to be of different races.
Twins can *be* of different racial mixes.
https://en.wikipedia.org/wiki/Superfecundation#Heteropaternal_superfecundation
So what’s the best way of separating out things that are genetic but also cultural? I assume various researchers are doing this (?)
I express my confusion here: http://lesswrong.com/r/discussion/lw/nf0/genetic_nature_is_cultural_too/
> That would make the quest to change important outcomes like intelligence, personality, income, or criminality by changing society even more daunting.
I don’t think this conclusion can be drawn because we seem to miss some essential relationships between parent-child interaction and society-citizen interaction. There are social programs that are very effective at influencing at least some dimensions, e.g. crime prevention:
http://evidencebasedprograms.org/about/crimeviolence-protection
And also consider all the biodeterminist parenting effects on IQ:
http://squid314.livejournal.com/346391.html
Out of that list, intelligence and personality aren’t gonna change much.
But a social system properly structured can absolutely change income and criminality.
This is easily visible just in the massive rise in violent crime starting in the ’60s, and the drop since the early ’90s. The people didn’t change, the social rules did. If we used to have less crime, then we are capable of getting there again. Of course, the problem is, no one actually knows how to change society in a reliable way without unintended consequences. But it is possible.
>Of course, the problem is, no one actually knows how to change society in a reliable way without unintended consequences. But it is possible.
The obvious solution is to invest in the development a superintelligent AGI that can figure this stuff out way better than we can.
(I’m only 2/3rds joking.)
3f. Your brain appears to have a particularly high degree of somatic genetic diversity.
http://www.theatlantic.com/science/archive/2015/10/the-genealogy-of-your-brain/408232/
I could speculate on a functional reason for this similar to the somatic genetic diversity in your immune system but the extra high mutation rate in your brain cells is clearly a source of non-shared environment.
I just always figured that “Non-shared environment” was dominated by things like “Got a crappy boss” and that schools and peer groups were part of the “Shared environment” for twins that remained in the same household.
Might there be a genetic component to being racist? And to what sort of racist?
I would guess so. I wasn’t able to find anything about the heritability of racism specifically, but I was able to find a study about the heritability of homophobia, and it’s fairly heritable.
I would assume (though I could of course be wrong) that racism is roughly as heritable as homophobia.
Smart alec question: what happens if the genes for homophobia/racism are more dominant/common in one race than another?
Then that race is more racist/homophobic than usual? Seems obvious.
I have a notion that people have lived under three basic sorts of society, and each might select for different genes.
Hunter-gatherer: moderately flexible behavior, egalitarianism, small group dynamics.
Primitive agriculture: hierarchy, tolerance of boredom
City: tolerance of living with strangers, disease resistance
Hypothetically, separatists, racists, and such don’t have a lot of city genes. I do think people who have a strong preference for dealing with those who have shared genes and culture aren’t necessarily the same a racists. “White”, “black”, “asian”, etc. are much too large and diverse a groupings to be innate preferences– races are partly theories and partly vague political coalitions.
Agreed. Classic example MAOA-2r variant. Not very useful for getting a full time offer out of your internship.
Well, homophobia confers a genetic advantage, since it enables your society to pressure people to engage in reproduction even if they aren’t inclined that way, and differential fertility is the driving force of evolution.
Or it confers a disadvantage because your society will be at Malthusian limits, and won’t enjoy the benefits of having more wealth in fewer hands…such as more education leading to better technology leading to more wealth…
Evolution doesn’t even notice your education. All evolution cares about is differential fertility.
assortative mating for homophobic attitudes
CONTENT WARNING FOR SLURS:
Pardon me while I go bang my head off the wall for a bit to see if that will make that phrase make more (or indeed any) sense in any fashion other than “currently in the social sciences political theory trumps hard fact and for our next paper we will protest about hospitals using evidence-based medicine after we’ve finishing bashing the queer-bashers and signalling how right-on we are”.
“Darling, want to go burn some faggots (haw, haw!) on the pyre?”
“Oh darling, the perfect date!”
Now, perhaps OKCupid has a new section for “Are you a homophobe? We’ve got the match for you!” but I don’t know how this is supposed to work in practice.
Is this a more complicated way of saying “straight people tend to marry straight people and if, perchance, they marry a gay or lesbian partner who is a victim of social pressure and is trying to ‘cure’ themselves, the marriage is likely to end in divorce (even if amicable on both sides) and if they enter a second marriage the straight partner is probably even more likely to try and be sure their spouse is straight this time round”?
Does that count as “homophobia” – well, in these progressive days, probably it does.
Re: Sastan’s question – what, precisely, would be the genes for homophobia? Is that specifically “ugh, I hate, detest and fear queers, I think they are yucky icky” or is it a constellation of attitudes about “I like clingin’ bitterly to mah guns and mah God but I’m agin’ the govinmint and gays”?
I ask as an ignorant conservative rural type, so I don’t have the larnin’ of my betters to know this answer by instinct. Seeing as how there isn’t a consensus on there actually being such a thing as “the gay gene”, I’m kind of interested to know if there is such a thing as “the anti-gay gene”.
How about: two people meet a conservative-denomination church group, strike it up, and get married? That kind of thing happens all the time.
Do you really think it’s unlikely that people who are members of conservative denominations are more likely to have a reflexive disgust toward transgressive groups like homosexuals, transsexuals, etc.? I don’t. Because if you’re sitting there hearing some sermon talking about the evils of homosexuality while thinking “Come on, I don’t see anything wrong with this,” maybe next time you go to the Episcopalian Church.
That’s not even saying everyone in the Southern Baptist Church is a fire-breathing homophobe. Maybe they find them 1% more disgusting than the general population finds them. If it is somehow genetic, then breeding together two people who each find homosexuals 1% more disgusting should, over time, result in children who find them 2% or 3% more disgusting.
But seriously, all your posts in this regard seem to be rants against “city-folk privilege”, which they are not “checking.” How could someone even dare to suggest something unflattering about social conservatives? It’s the same kind of idea as saying white scientists can’t investigate the question of racial differences in IQ because of white privilege and haven’t you ever heard about eugenics.
> burn some faggots
I remember when I was in 7th grade a schoolmate called me a faggot. The only meaning I knew for that word was the “firewood” one used in the tales of the Brothers Grimm.
[That’s me in a nutshell- bookish and not exactly with-it.]
“I remember when I was in 7th grade a schoolmate called me a faggot. The only meaning I knew for that word was the “firewood” one used in the tales of the Brothers Grimm.”
When I was in summer camp some of the other boys asked me if I knew what “intercourse” was. Of course I did. Going between, communication, stuff like that.
Pretty sure I knew the facts of reproduction at the time, but it didn’t occur to me that that was the sort of intercourse they were asking about.
Hypothetically, homophobia could be related to having a vivid imagination about sex.
Very homophobic man sees or thinks about a gay man, and immediately imagines having sex with that man, and “knows” either that they would hate it a lot or that they’d like it which would put them in danger in a homophobic culture.
Someone with a weaker imagination (in general or about that particular subject) doesn’t have the fast personalized reaction.
What you imagine vividly strikes me as something which could be genetic.
No wonder there is no consensus on there actually being such a thing as “the gay gene”, since the way these things work means there is virtually no chance that such a thing as the “gay gene” exists.
Go talk to Greg Cochran or Jayman if you want more detail.
That’s really interesting. Is tit more likely that this is its own thing or that we’re just seeing a heritable “disgust reaction to things that seem unnatural” trait? (I know very little about heritability, so pardon my ignorance if this question is nonsensical or obvious.)
http://humanvarieties.org/2015/11/01/heritability-of-racial-and-ethnic-pride-preference-and-prejudice/
Three points (sorry, not reading through the comments, so maybe already discussed):
(1) Regarding your title, aren’t schools and peer groups part of *shared* environment? Kids from the same family probably attend the same (or similar) schools and have the same (or similar) friends. And this is much much more true of identical twins, who probably literally have the same social circle their entire childhood.
(2) Unshared environment is basically a residual. Until we identify a specific thing in “unshared environment”, standard statistical practice would be to just call it noise. So it would probably be better to look at the share of explained variance that is explained by genetics (close to 100%), and say that nature wins in a landslide over nurture, “but time and chance happeneth to all.”
(3) I read somewhere (I think in the terribly-named book “Intelligence: all that matters”) that the correlation between multiple attempts on an IQ tests by the same person is about 0.80. Coincidentally, this is *also* the correlation between the IQs of identical twins. I *assume* that the latter refers to IQs averaged over many attempts, because otherwise it suggests that identical twins just have the same IQ, full stop. In any case, it puts the degree of non-genetic influence on IQ into perspective. (And is hard to square with the Flynn effect, etc etc).
Shared environment is typically studied by comparing unrelated individuals raised in the same household. Kids of different ages probably don’t have the same peers.
Two quick comments, first on your second point “2. Luck of the draw” I doubt that you can strengthen the distinction between luck and non-shared environment enough to build an argument out of it. To me the example of different CEOs sounds like the definition of non-shared environment factors.
More importantly on your first critique “1. Error,” that argument can potentially cut both ways, consider both twins toss a coin, you will get 50% correlated (head-head, tails-tails) and 50% uncorrelated (head-tails), which suggests to me, that some kind of auto-correlation of the error should end up as an error for shared factors and the rest of the error should end up as non-shared. Which is somewhat unsettling since it sounds like it would break a lot of the assumptions of standard statistics, for example the error is certainly not Gaussian.
Furthermore the, to me, very surprising result that shared environmental factors are negligible, even when non-shared are not. Consider as illustration the talent for playing football for American Samoans compared to neighboring islands, the environmental factors are important if the twins are on different islands but not when they are on the same island. This sounds like non-shared environment factors are simply easier to measure than shared factors. So I would read the 50/50 result for shared and non-shared plus genetics, as mostly technical difficulties. ( To be clear, I don’t think that the researchers do something wrong, I just think that the task is very hard.)
I agree. There is strong evidence of environmental factors producing large effects in the short-term. I suspect that causes a lot of the effects of nonshared environment. Even if shared environment doesn’t have much of an effect on two people when they’re 30, what they did yesterday definitely does. There is a very clear cause in that situation; the decisions of the two CEOs. Ascribing something to luck denies determinism.
5. “May you live in interesting times” theory:
Two hundred years ago, two twins would likely be very equivalent farmers, and sire equivalent number of children. Except perhaps for childhood maiming or disease, e.g. Helen Keller / Johnny Tremain. That was clearly far more common source of variation in the past (or maybe not because the injured ended up dead, not handicapped ?) but there also seems like there was social pressure to get out there with your peg leg and earn your keep.
Sixty years ago, soviet twins might face very different fates based on where they were deployed in the war, and whatever other political vagaries they had to endure through the Stalin years. But with 10% of male population eliminated, to the survivors go the spoils. Again, historically, there seems to be more exogenous variation.
What is nature of life outcomes today: Is it different in the ghetto than the ‘burbs? Different from the Shetl of yore? Just as 6’0″ is a giant in 1500’s France, I would say nature/nurture is not a universal split, but has at least 80-to-20 swings based on time-and-place.
Is the 50%-50% idea itself confusing the matter? A model that I think is more helpful is asking “which side of a rectangle has more influence over the area?”
In terms of a rectangle, the 50% idea only works if you have a square. Then, in some cases, one side of the rectangle has more influence, and in other cases the other side has more influence. But in all cases, the relationship is the same, multiplication.
It seems to me to be the same here. In some cases, being born with a low IQ or some other disability has a huge influence on the outcome. In other cases, being born in a low income family, neighborhood, or country has the huge influence. And in other cases still, the two are balanced against each other.
The Nature side of the equation is also overcounted.
1) Prenatal Environment
Twin-studies lump in prenatal environment with Nature, because twins (by definition) share a prenatal environment. That may seem fitting, since the prenatal environment cannot be changed after the fact. However prenatal environment should be on the Nuture side of the ledger, because in can be changed (for future generations) by intervention into the Nuture. In this way, its not so different from childhood development, which is clearly Nuture. The Nature-Nuture debate can be used to assess the effectiveness of reducing poverty, and this is an important effect. In a country like the US that doesn’t have universal equal healthcare, wealth and financial security may be a significant factor in prenatal environment. There may be effects from prenatal nutrition, effects from prenatal pathogens, and effects from the health of the mother (e.g. stress).
2) Effects with depend on Nature & Nuture
It is well-known that black children adopted by white parents do not escape the detrimental effects of racism. These detrimental effects are hereditary, in that black skin is hereditary, but a racist environment is a necessary component as well. If we could eliminate the racist environment, black skin would cease to lead to detrimental effects, so we should really classify these effects as a result of Nuture. Again we use the criteria that Nature is what cannot be modified by intervention into the environment. When talking about to what extent something like Intelligence or Criminality is caused by Nature, we should only count the subset of genetics that necessarily contribute to the outcome not the subset of genetics that conditionally contribute.
One cannot design a twin study in which one black twin is raised in an environment completely devoid of racism. Even by looking at a twin raised in a less racist town vs. a twin raised in a more racist town, the whole effect cannot be captured when one cannot extrapolate down to zero racism (and also no one lives their life solely in one town). One can hold race constant and look for hereditary variation within each race. But even there, the degree of “blackness” will contribute. Individuals who are deemed to “look black” – skin, hair, facial features – will experience more significant racist effects. The genes for those traits will be erroneously marked as genes for other effects.
Taking it away from race, we can imagine there are many traits like this. Some white people “look smart” or “look honest” relative to other white people. Its well know that height (and attractiveness in general) creates a bias towards success. Supposed know that height contributes to IQ scores, height is highly hereditable, and Intelligence (as measured by IQ tests) is heritable. These results fight with each other. Some fraction of those genes that are found to contribute to high scores on IQ tests are not actually from intelligence, they are actually for height which then translates into a better educational environment.
There are some things in this category that are admittedly difficult to classify as either Nature & Nuture. Let me use myself and my identical twin as an example. Let’s say that we are both intrinsically intelligent, we are both intrinsically extroverted, and we are both intrinsically deep-focusing. We both did well in school because of all three of these things, and if we were raised in different environments we would still do well in school. If we use performance at school as a proxy for intelligence, however, we would mark all three of these factors as part of the Nature contribution to intelligence. However the last two factors are conditional. If we got more educational attention because of our extroversion, that is an outcome that is conditional upon the cultural norms of our society. And because we are deep-focusing both of us test well. Because we are deep-focusing. both of us would have done worse if class times were shorter and better if class times we longer. If we are both raised in the US, those factors are unlikely to change no matter where we were raised. But yet they are subject to cultural intervention, if say one of us were to go to a specialized private school or there was a large-scale educational reform. But maybe we should also say that deep-focusing should also be consider a component of intelligence. If we say that True Intelligence is related to performance at deep-focusing tasks, than the schools are right to prepare students and rank-students on the basis of deep-focusing tasks.
1. The womb is counted as shared environment, since both MZ twins *and* DZ twins share a womb. The only case in which womb is not counted as shared environment is insofar as MZ twins have a more similar uterine environment than DZ twins, which I think happens in certain cases involving a structure called the chorion that my medical school professors would be horrified I don’t remember anything about. But I think most intrauterine issues are in shared environment.
2. I think twin studies would not count gene x racism interactions as genetic, since all twin pairs are concordant with respect to race.
I assume that these studies only look at twins? Because if you examine siblings who didn’t share a womb, it’s obvious that the uterine environment might change between pregnancies and so be part of “unshared environment,” even without taking into account things like the Older Brother Effect. For example, a woman might drink during one pregnancy but not the other, or live in an apartment with flaking lead paint during one pregnancy and then get a better job and move out before getting pregnant again.
Uterine environment stuff is not important. We know this because of this study
http://link.springer.com/article/10.1007/s10519-015-9745-3
and because ordinary siblings are about as similar as DZ twins who did share the uterus at the same time. They have the same genetic relatedness of .50 by descent.
First, the paper that you linked doesn’t seem to be as strong as you claim. I quote from page 7: “Intra-uterine environmental factors do influence the intra-pair similarity of MCMZ and DCMZ pairs for birth weight, weight during the first years of life, achieving motor milestone standing alone, externalizing behaviors at age 3, internalizing behaviors and anxiety at age 12, and autistic behavior at age 7.” I’d be very reluctant to wave all of that away as unimportant.
The abstract claims that these results would only have “small” effects on heritability estimates, but the numbers brought up in the Discussion don’t sound that small to me: “For internalizing behavior at age 12 years, with a difference of 0.11, the heritability will be 50 % [2*(0.71–0.46)] when including both types of twins and 38 % [2*(0.65–0.46)] when including only DC twins. For anxiety at age 12, the heritability will be 70 % [2*(0.71–0.36)] when both types were included and 52 % [2*(0.62–0.36)] when only DCMCs were included.” Maybe these results are due to chance and as such these effects aren’t real, but we’d need to look at more studies to determine whether or not that is the case.
Regardless, even if we were to assume that research generally finds uterine environment to be unimportant, isn’t the (as far as I can tell, let me know if there is something that I missed) very well established Older Brother Effect – that is, having more biological older brothers makes a man more likely to be gay – a glaring exception to that? And if it is, doesn’t that raise the question of what other effects we might have missed?
Here, I did the analysis for you.
http://emilkirkegaard.dk/en/?p=5873
What you wrote makes sense. Thank you for explaining.
The fortune cookie from my lunch today at a Chinese restaurant had input relevant to this thread.
Progressives taking over everything…
There’s no good reason to ignore factors that are intermittent:
It’s amazing how little influence environmental factors have, once you define away a large group of them. Or, what amounts to the same thing, define the outcome of interest not as performance, but some stable underlying factor (i.e. intelligence).
Right. IMO, almost every example Scott gives is, in a literal sense, a legitimate issue of non-shared environment. One crack dealer has an undercover cop in their neighborhood, one doesn’t. That IS non-shared environment and DOES explain why one gets arrested and one doesn’t. Same thing with inhaling cat dander or whatever.
It sounds like Scott’s main issue is with people who understand “non-shared environment” to exclusively mean inherent personality differences at the complete and total exclusion of luck and other random factors. But I personally never saw it that way.
I’ve mentioned this before, but these studies completely ignore what happens in the womb. Not all of what happens in the womb is genetic. If a low IQ mother drank while she was pregnant, had twins whom were separated at birth, and the twins had a low IQ, a twin study would conclude it is evident that the mother’s low IQ contributed genetically to her children’s low IQ, completely ignoring the fact that the mother drank while pregnant. That’s an extreme example, but there are a ton of poor life outcomes associated with a low IQ (namely, income, which is associated with smoking, drinking, lifespan, diet, illness, weight, etc.), all of which could theoretically contribute to a child’s development while in the womb. It’s not easy to study (surrogacy? Would be so hard to get a large sample size), but we shouldn’t ignore it, either.
At first I thought that, while this could obviously be a concern for adoption studies, in twin studies it would show up as either shared or unshared environment. But then I Googled “fetal alcohol syndrome twins” and found this study:
(emphasis added)
So yes, it looks like this could definitely be a problem for twin studies, especially if other uterine environment conditions like smoking while pregnant exhibit similar gene-environment interaction.
Most twin studies are based on comparisons between MZ and DZ twins, both of which share the same womb. This means that whatever happens in the womb is attributed to environment, either shared environment (if both twins are equally affected) or unshared environment (if they are differentially affected).
Unless DZ twins are more differentially affected than MZ twins due to gene-environment interaction, which does seem to happen with fetal alcohol syndrome (see my comment directly above). In that case, unless I’m misunderstanding the study design, some of the measured genetic variation in whatever is being studied may actually be genetic variation in resistance to alcohol poisoning or other pre-natal environment conditions.
That’s possible, but the FAS study had 5 MZ and 11 DZ pairs, which means that it lacks statistical power to support any conclusion.
If you want to know – “what will be the effect on Y of changing X?” then you need to perform the experiment. Knowing how much of the variability in Y is explained by the variability in X, only gives you loosely related information. (It can be very useful for generating hypotheses, but not great for actually proposing policy).
One of the better education interventions was the Perry Preschool project. Not bad results for a 1 year part time (2.5 hours every weekday) intervention in my opinion.
Twin studies are experiments of nature. Randomization happens because of Mendel’s laws and the random nature of twinning.
I’m not impressed by the results of the Perry project. For example, 36% of the treatment group had been arrest 5+ times by age 40, compared to 55% of the control group. That’s a better outcome, but still awful. Morever, there are serious questions about the validity of the Perry project. It’s an old study with a tiny sample size and, contrary to what is usually claimed, assignment to the groups wasn’t random.
Yes twin studies are fine for studying genetics, that’s not what I’m addressing – if you want to study environmental interventions then study them directly.
This paper claims to statistically adjust for the compromised random selection in the Perry project and find that the effects are still significant. I’m not good enough at math to tell if they are right.
Even if that adjustment works, which is debatable, what I find ridiculous is how this one small, old study has been turned into this epitome of social science research which is cited and reanalyzed ad nauseam and used as justification for wide-ranging policy changes. Why, especially with the revelations of extremely poor reproducibility of psychological research in recent years, do people pay so much attention to this study? In a saner world, it would be dismissed unless its results could be repeated in a large-scale replication. Early on, the Perry researchers turned the program into a commercial operation which in itself should make us skeptical of their results.
Perhaps one of the reasons that it is cited so much is precisely because it is so old that we have extremely long term data on its subjects. How many social science studies are there that followed their subjects from pre-school to age 40? I’m not too familiar with the field, but I imagine that the number isn’t very large.
Sure, but given that the Perry study has been hyped for several decades now (see Herman Spitz’s 1986 book The Raising of Intelligence for a critical review), you’d think that someone would have tried to redo it. Even a shorter follow-up to early adulthood would be very informative.
“At least in relatively homogeneous samples (eg not split among the very rich and the very poor) studies of many different traits tend to find that ~50% of the variation is heritable and ~50% is due to non-shared environment, with the contribution of shared environment usually lower and often negligible. This is typically summarized as “50% nature, 50% nurture”. That summary is wrong.”
On the website of the meta-analysis of all twin studies ever (up to a few years ago), looking at the average reported ACE estimates I see:
0.468 for genes (h^2)
0.177 for shared environment (c^2)
0.355 for non-shared environment (estimated by subtracting the other two from 1).
That’s not 50-50-0 (although adult IQ has a smaller relative shared environment contribution than a lot of other cognitive traits, including IQ at younger ages). There’s room for meaningful environmental effects in there, especially if the large effect exposures are rare (even a big effect that only covers a few percent of the population won’t contribute that much variance).
For ‘major life areas,’ which includes education, relationships, etc, the average reported ACE figures are 0.297, 0.277, and (inferred) 0.426.
http://match.ctglab.nl/#/specific/plot1
Fixing measurement error will boost shared environment effects alongside genetic effects.
Another thing to think about: these variance components are coming from squared correlations. Take the square root to get the correlation coefficient. If common environment accounts for 10% of variance on a trait, then 1 standard deviation of environment can get you a little less than 1/3rd of a standard deviation of the trait. If genes account for 50% of the variance, then 1 standard deviation of genetic factors will buy you a little more than 2/3rds of a standard deviation of the trait.
You talked about this in another post:
https://slatestarcodex.com/2015/08/02/stalin-and-summary-statistics/
It seems when people say ‘epigenetics’ it’s usually code for ‘I’m depressed because my mom had a bad childhood and that totally can be inherited because epigenetics’.