Sentences to ponder

On average, students in 2014 in every income bracket outscored students in a lower bracket on every section of the test, according to calculations from the National Center for Fair & Open Testing (also known as FairTest), using data provided by the College Board, which administers the test.

Students from the wealthiest families outscored those from the poorest by just shy of 400 points.

From Josh Zumbrun, there is more here.


Interesting that this is posted as the same time as

The word you're looking for is "deprivation," not "inequality." If we removed all students except those in the lowest income bracket, the latter's test scores aren't going to increase. Thus it has nothing to do with "inequality."

@Ricardo: I like your point in general, but in this particular case at least, it could very plausibly be argued that while the scores of the poor students will not increase, these students are now more likely to make it to top colleges. Unless you want to assume that these colleges will choose to keep those spots unfilled, which is not very plausible...

Fair point!

And graduation rates at top colleges will decrease.

Not if we threaten them with lawsuits!

Sounds like Pol Pot may have a solution for us then.

I dont agree. If colleges cannot produce degree holders with the skills and abilities desired by the work force, then the degrees will become meaningless and hence colleges WILL disappear.

Eliminating the top 80% is a useful thought tool, but remember that this is not possible. Nor is 80% of the future workforce going to disappear and landscapers will become tenured professors, lawyers, and doctors. Reminds me of Idiocracy.

A better argument is that the resources of education are not allocated to produce equal outcomes. NB I dont consider this a good argument, just a plausible one.

Nope. This is more of what I had in mind:

Let me guess. This is proof that the test is unfair?

Or is it proof that we need more welfare handouts?

Pretty sure those are the only two possibilities.

You have a very limited imagination.

Pretty sure you're missing "The Other Jim's" sarcasm. It seems to me that the most likely explanations are education, both in terms of the parents and better schools for the kids, and genetics (IQ, deferred gratification, etc.).

Perhaps this community is not rich, people here seem blissfully unaware of the extremes in modern "test prep." I know families who drop thousands on tutoring, and still hire a psychologist to say that their precious child needs extra time on the test.

Go ahead, plot extra time vs parent's income.

Go ahead, plot extra time vs parent’s income.

In America, income correlates very strongly with hours worked.

I have no doubt that some parents dump thousands, or even tens of thousands, on test prep. This doesn't mean it works.

Or you could have just read the linked article where it talks about test prep.

I am aware of test prep training, but it is my understanding that on average it doesn't boost a students scores that much.

It is pretty crude to plot so many levels of income, and then only consider one level of "prep," isn't it?

The rich people are hoarding all the free prior year standardized testing on the internet, which I'd accessible by smart phones. It's obvious.

It's also pretty crude to imagine test prep as some mysterious good that explains all test variation. I know a free test prep called public schools, why do poor kids choose truancy over their free test prep, and why do you think they'd show up after class to tutoring.

There are a whole raft of advantages which increase with household income.

1) Reduced violence in schools means reduced stress, greater concentration.
2) Reduced violence in homes means reduced stress, greater concentration.
3) Increased respect for education means increased educational support
4) Increased wealth means increased educational resources (including test prep)

I just chose test prep off the top because it seems so outlandish these days. Rich parents go nuts on test prep. I saw a "test prep consultant" who charged parents a few thousand without even doing the prep, just *planning* the prep.


Did you read the recent study that taking a poor family and making it well-off does not improve outcomes for siblings born after the transition?

Money, in itself, doesn't cure any of my list, 1-4. You need a secure environment with a education-valuing family/culture. That is harder to build.

Proof that smart unborns chose rich parents, and stupid unborns chose parents in poverty.

After all, the people who are poor and stupid chose to be poor and stupid, and the rich and genius chose to be rich and genius.

I have met rich people (people in the top 2 quintiles) I considered below average intelligence, but very few (and I bet quite number of those were the children of rich people, though I don't know it for certain)- most I have known were above average in intelligence, and certainly above average in ambition. I have certainly known people who were above average in intelligence that were quite poor, but they all, with not a single exception, had a serious substance abuse problem and/or a mental illness.

I think you are far too quick to dismiss the idea that poverty is largely a consequence of inherited traits. I think the ways we treat poverty probably isn't optimal because of this politically correct attitude you display.

I believe it was reported in these pages that genomic studies show only the mildest link.

But you know, don't let that upset your priors.

Cite please?

I believe you're referring to a study that showed that no *single* gene variant had outsized impact on overall intelligence. That says nothing about the proportion of variance attributable in aggregate to genetics. For example it's well known that in the developed world nearly all explainable variance related to height is genetically determined. Yet the same genetic analysis when applied to height shows that the contribution of any single gene variant is well less than 1% of observed standard deviation.

Similarly with intelligence, twin studies have revealed that virtually all explainable variance is genetically attributable. The reason height and intelligence can be nearly 100% genetically determined, yet still have no single gene variants that strongly contribute, is because the phenotype is multifactorial. I.e. a very large number of genes (most likely in the thousands) have an effect on height or intelligence. So you're not going to find any single gene that's highly correlated with being smart or tall.

I think you are drawing a "hope" not supported by the data, Doug. The "effects are maddeningly small" and it is not a case that a large genetic component was found which then could not just be mapped to particular genes.

The height thing is funny because we all know that our relatives back in Europe (or Asia) were about a foot shorter than we are, right?

"When a given environment maximizes the genetic potential of a population for a given trait, this population tends to have a higher heritability for that trait, and vice versa. In developed countries, nutrition for childhood development is strong, which maximizes the genetic potential for height assuming no selection or new mutations. Thus, the overall heritability estimates tend to be higher, i.e., 80 percent. In contrast, in developing countries, nutrition deficits lead to a lower heritability. The fact that the mean height of the U.S. population has almost plateaued in the past decade suggests that the nutrient environment has almost maximized the genetic potential of height, at least in this country."

(Repeated from my comments on that post)

Identifying individual gene variants at the statistically significant is likely to take a very large sample size. The alleles they did find at this sample size are likely to be the largest in magnitude, and represent an upper bound of contributing 0.3 IQ points. At a standard deviation of 15 IQ points, that implies a minimum of 2500 individual alleles, assuming that their distribution is uncorrelated. A gene variant that contributes 0.03 IQ points would require a million person sample size.

So, if we give every kid a healthy diet and a low stress, highly supportive, educational environment, their outcomes might plateau as well.

We are a long way from that.

"Genome-wide association studies establish that human intelligence is highly heritable and polygenic" was published in 2011, and the full text is freely available.

For those who might want to read the paper, its brevity is inviting.

So they're simultaneously well fed and unstressed when it comes to growing, but underfed and stressed when it comes to brain development, John?

Cornflour, a study which says 40-50% of the variation comes from genetics is not the same as one that says 40-50% of the intelligence does. We already know that the stars need to align for a top 1% (or 0.1%) IQ. In the middle ground, the body of this income/SAT graph, a lot is going on. Certainly this paper was not able to show how many points (in IQ) that any group lost or gained. As the paper says, looking at one individual, you'd never find a signal in the noise.

Careless, different groups, different generations.


I cited Davies et al. as an example of the many GWAS papers supporting a significant link between genetics and intelligence, which you dismissed as mild. The paper that you cited in support of your dismissal ("Smart genes prove elusive") is a news article summarizing an editorial policy statement and two papers by the same research group. I've just read all three. I'm not trained as a geneticist, and this is not my field of expertise, but nothing I've read leads me to conclude that the dispute over the role of genetics in intelligence is anywhere close to being resolved. Simply dismissing the link between intelligence and genetics seems unwise, and rude remarks about someone else's priors are regrettable.


Endless amounts of twin and adoption studies have shown that 40-50% of intelligence and personality is explained by genetic inheritance, less than 5% from family environment inheritance, and 45-55% is unexplained variance. Meaning that if two random people have a 100 mean squared IQ difference, then two identical twins will have a 50 mean squared IQ diff, two twins raised apart a 55 mean squared diff, and two adoptees into the same family a 95 mean squared diff. You seem to be convinced that the residual 50 squared point variance still found even between identical twins is attributable to some mysterious X controllable factor: nutrition, stress, cultural attitudes towards education, etc.

In all likelihood it is simply noise. An early childhood infection at an inopportune growth period, a cosmic ray in the womb, or even maybe the first twin didn't get a good night's sleep before the test. It is extremely rare in the social sciences to find any model that explains more than 50% of any social variation. If the unexplained variance is some sort of developmental X factor, we have to justify why children raised by extremely resourceful and educated parents seem to be just as likely deprived of it as impoverished ghetto kids. Furthermore what hope does social policy have of adjusting this X factor, when we don't even have any clue what it is. If even wealthy parents with direct control over their kids can't control the X factor, why would anyone else be able to?

Careless, different groups, different generations.

huh? Do you think that the poor suddenly got really short recently? They're still the same general height. No evidence of malnutrition.


How do you know we're a long way from providing enough of those things that genetic variation will explain most of the IQ or test score variance?

How about school funding formulas which are propitious to similar levels of opportunity for children, as opposed to current formulas which essentially guarantee the above results.

Are rich people so insecure that they cannot compete with children of the poor without stacking the deck?

As a "rich" and upper IQ person who happily paid hundreds of thousands of dollars for my kids education starting with kindergarden, competitive insecurity has nothing to do with it. We gave our kids the best possible start in life that we could afford. No McMansions, no $100,000 cars for us or them, just love, ethics, charity, manners, travel experiences and education.

Not giving is the same as taking. /derp

I think school funding is low on the list of drivers of a kid doing well.

Are rich people so insecure that they cannot compete with children of the poor without stacking the deck?

Yeah, what's with those damned rich people, trying to give their children the best opportunities?

If school funding worked, then absolutely yes I would be 100% in favor, it would be a good investment. But what evidence is there of this? I think there is better evidence for school reform.

Nothing against trying to give your kids personally a better chance. Just against systems which entrench it.

How about compare average funding to average income in different areas and then pretend that funding, not income, explains results?

This lib is unintentionally funny:

How about compare average funding to average income in different areas and then pretend that funding, not income, explains results?

How about we pretend that I am right and you are wrong?

To me this sounds like we don't need the college board. If parental income correlates this strongly just use parental income for acceptance. Pick the top percentage out of each group. Done.

The size of the correlation (which is just the square root of the r-square) is not stated in the article, and cannot be calculated from the information given. The idea was to show a strong relationship, but it fails to do that.

Essentially they are only comparing averages of each income bracket it seems?

I don't know if you're joking or not. This doesn't study doesn't tell you how much variation you get within income groups compared to across income groups. The left has been pretty effective at persuading people to not even think about this aspect of the issue.

Yeah, in a previous iteration of this argument, confused reporters (Catherine Rampell) were calculating the correlation coefficient between the income levels and the income group *means*, thus removing all the within-group variation. Kind of funny and kind of sad.

Ack! Looks the the drive to make reporting look "data-driven" is yielding its first casualties.

This is the reason that capitalism in practice doesn't look that different than feudalism. Usually the children of the aristocrats really do tend to be more intelligent and talented than the peasants. Carefully combing the entire populace for high-level workers is often superfluous. There's not that many diamonds in the rough. Someone working with the assumption of blank slate, no genetic impact, model vastly overestimates the economic benefits that come from meritocracies. That doesn't mean there's not valid fairness arguments, but the efficiency arguments are weak.

In capitalism, the aristocrats work more.

Sure, but how much of the work is largely unnecessary and just used to sort people in a way that's not directly tied to their last name? Particularly the activities that generally fall under "paying your dues." Investment banking associates working 80 hour weeks, mostly tweaking powerpoint presentations. Aspiring doctors cramming organic chemistry and physics. Children of company founders working starting low-level jobs in the firm to maintain the veil of meritocracy.

On the flip side, from Isaac Newton to Linus Torvalds there's a long tradition of affluent children accomplishing amazing things while pursuing independent interests. A society that highly values the ideals of equality of opportunity will have less tolerance for these indulgences. Today even the children of the very wealthy are culturally, and to a lesser extend economically, pressured to pursue a standard career path. Bill Gate's kids would probably have a higher contribution to humanity by using their massive genetic advantages and inherited wealth to read, write and create on completely independent terms. Instead they'll almost assuredly hold conventional jobs for good portions of their lives.

"Children of company founders working starting low-level jobs in the firm to maintain the veil of meritocracy." This is often done not to maintain a veil of meritocracy but to teach the kids what the business really does. If you run Waste management but have never ridden a garbage truck I think your understanding of the business is likely to be pretty flawed. (see K-mart)

"Children of company founders working starting low-level jobs in the firm to maintain the veil of meritocracy."

That is not what I saw. What I saw was effort of the parents to expose the heirs to all aspects of their future business. In the same way that a future military officer still needs to know how to fire a gun, even if the grunts normally do that.

The agent/principal problem is very strong when it comes to employing of various MBA'd managers. An owner / stakeholder who does not have a clue about his business falls easily to the PowerPoint smoke-and-mirrors show.

Actually, capitalism in practice looks *nothing* like feudalism, as long as you're observing over time.

In Feudalism, there's no way for the peasants to become aristocrats or ways for the aristocrats to become peasants. But within capitalistic systems, you see movements of that kind *within* generations.

Stop trying to justify some sort of faux-nostalgic vision you hold via gross distortions of the way capitalist systems *actually* work.

Feudalism is filled with examples of people moving up and down the chain. Even to the level of sovereign. Where'd the Swedish royal dynasty come from? A low-born French peasant, who proved an astounding general in the Napoleonic Wars. Where'd the House of Liechtenstein come from? A minor dynasty that shrewdly acquired so much land and influence over the generations, that the Holy Roman Emperor was compelled to elevate them to the level of principality? The Tudors were nothing more than illegitimate issue who had no legal claim. But Henry VII was politically smart enough to build an alliance, militarily gifted enough to defeat Richard III, and foresightful enough to rebuild the treasury, consolidating his dynasty's hold on power.

The difference between feudalism and meritocracy is that the latter considers the low-born and high-born equal until proven otherwise, whereas the former does the inverse. In reality a large majority of the time the high-born are more competent and talented, so the results in terms of mobility largely end up the same. That's why Greg Clark (link at bottom) that the rates of intergenerational social mobility are no different in feudal Europe than they are in modern Europe. Furthermore he found that a very disproportionate number of the social elite today are descended from the aristocratic families of prior centuries.

Feudalism in the European and Japanese Middle Ages was less rigid than you seem to think.

There is plenty of things to hate about feudalism, but no society exposed to constant (unfriendly) contact with various neighbours would survive long if it did not have some meritocratic mechanisms built-in.

As a low-born, you wouldn't probably become a king, but you might aggregate enough fortune and influence that your position in the court would be significant. At that stage, the royals would probably bestow a hereditary title upon you (unless you were a non-Christian).

Doesn't look that different than feudalism?? Wasn't serfdom, i.e. slavery, the defining characteristics of feudalism? Along with distributed power centers? None of which we have?

Legal restrictions associated with slavery and serfdom were not the marginal constraining factors. Were there some who were held back from opportunities by their legal status? Of course. But if 12th century France had everyone sit for the SATs to assign career paths, the vast majority of peasants' children would place as peasants. And the vast majority of placed aristocrats would be the offspring of aristocrats. We know this because Gregory Clark showed that systems with standardized test have nearly the exact intergenerational mobility as systems with inherited titles.

It might be that we don't even need the college. Educational institutions don't have much incentive to say that they produce good graduates because they choose the best in the first place. If at some (never to arrive) point we all stopped looking to university (or school) for kudos, we might realise that we don't any particular special school at all. (I'm not saying that we don't need education, but if it's all about the genes, then most places are probably fine).

"If parental income correlates this strongly just use parental income for acceptance. Pick the top percentage out of each group."
Trust me, the colleges are *way* ahead of you here

Surely, you're joking. Fairness requires that bright kids from poor families have access. The final caveat should have been the initial paragraph:

A final caveat: Within each income category, of course, is a tremendous amount of variation. There are students from wealthy families who do very badly and students from poor families who do very well. Having wealthy parents gives a leg up. But parental income is not destiny.

I can think of no greater incentive to be successful in business than the opportunity to facilitate your children's success. No yacht or African safari compares to having your kid get into a good school.

On the other hand, if your kid has reached teenage years and still fails to shine, you might as well blow his college fund on a yacht. At least he can borrow it and score some popularity points at school.

And he might marry a rich girl!

It is pretty awesome how the article assumes this is a result of income inequality without any even acknowledging that someone might dispute that point. As if compressing the income distribution would necessarily reduce the spread in scores.

It would be interesting to know how much the spread across incomes compares to the standard deviation of scores within an income bracket.

Well what are the other possible explanations as opposed to income inequality?

Any trait which is a) heritable, b) correlates with income, c) correlates with SAT scores
Intelligence and discipline are two obvious candidates.

Also cultural traits, while not genetically inherited, are candidates.

It's embarrassing that the author doesn't even raise the possibility that IQ is genetically transmissable.

This fixation on reforming the test until it gives us the results we want is crazy and potentially destructive of information. We should try to measure college readiness and cognitive ability, not just the portion of those traits that is uncorrelated with family SES. While the latter may be very useful, it seems better to just condition on (adjust by) SES if you want to achieve social justice objectives. This may seem politically naive, but we should really grow up as a culture and come to grips with the possibility of inherent (e.g. moral) tradeoffs involved in meritocracy, rather than desparately clinging to a fairy tale that "true merit" is uncorrelated with family SES. That is, we may want to trade off some meritocracy to get a better social outcome, and we should see that for what it is so that we can make decisions with open eyes.

I can't find the adoption study that compares the correlations between adopted children's "cognitive ability" and their biological vs adoptive parents SES. Adoptees' IQs depend greatly on biological SES, after controlling for traits of adopted parents. (Actually it's mostly some bar charts if I remember, stratifying by biological parent SES.) Below is a link to a little overview. But McGue, Bouchard, Plomin and Turkheimer are the psychologists / behavioral geneticists to look at on this topic (as far as I know).

SES - Socioeconomic Status

Yes, even if you're not prepared to entertain the possibility of inherited traits playing a role, how could anyone possibly overlook the fact that a)people of different social classes have different tastes, interests, habits of mind, etc. and b)that this wouldn't influence educational outcomes?

Wait, what? Are you serious? The idea that people who are successful will manage to raise successful kids does not even cross your mind at all?

The question here is what would happen if we increased the level of income transfers from the rich to the poor. Would these results change at all? I'm guessing very little. I doubt anyone would even try to disagree. That means you don't think inequality is the thing causing the spread.

Do you know about the Great Gatsby Curve?

Is this result novel? In other words, was there ever a time in American history where this sort of wealth-score hierarchy would not have held?

Ok, income not wealth.

Yeah. When was average not over again? When everybody was eating dirt? Good times.

New immigrants who came with nothing may have taken a couple of generations to make it into their proper place.

We have more immigrants now than we've had in a long time, don't we?

I don't know about the magnitude, but I'm sure these are big factors:
1) When there were more immigrants in the population who hadn't gone through the system yet. Even today, you will see that poor immigrants from Asian have very high-scoring children.
2) Before the advent of standardized testing, smart parents were less likely to have made it into the upper class, so society was less pre-sorted.

Yes but have poor immigrants from Asia with smart children ever been a cohort that'd be large enough to make a dent in the aggregate statistics?

Wealth/high income and work ethic is also highly correlated. Couldn't you just replace "wealth" with "work ethic" at come to the same conclusion?

How hard would you work for $1?

Opportunity matters.

How hard would you work for the opportunity to make more than $1?

Work matters

So, what if the chances you saw that seemed plausible were not as high opportunity?

Would you work harder or less hard depending on how much opportunity you perceived?

Do children with wealthy parents perceive themselves as having access to greater opportunity than parents of single moms in the hood?

Will equally motivated children work harder (or towards different goals) when they are aware of different levels of opportunity whether due to role models or connections?

Yes, blah, blah, blah, Nate, but as you know, and as these results prove: the difference between rich kids and poor kids (on average) isn't simply parental income.

The top performers are top performers because of hard work and discipline.

The monied can afford prep courses: one explanation.

The monied parents are more likley to force their kids to sit in on a school night and either study more or get rest for the next day's school work.

The poroer exhibit higher propensities to not push their kids to work hard, less likely to get extra help for a lagging pupil,

The difference, in addition to money, may be the higher emphasis and value more afluent people place on academic work and achievement.

"The monied can afford prep courses ..."

I've heard this forever, yet I also know public libraries are full of test-prep books, and even if one buys a book it's not that expensive.

To be fair, I used a cheap book instead of a prep course to prep for the SAT (and took the GRE with no prep). I acknowledge that the book probably requires higher motivation than the prep course, but I don't see how it's inferior.

For what does a prep course do, if not show students what types of questions will be on the test, and how best to budget time?

Even the linked article dismisses the "test prep" explanation.

Let's think about this. Some kids buy a prep book. Some students attend prep sessions with 30 other kids. And some students get 100 hours of one-on-one tutoring.

Does it really make sense, as in the linked article, to "average" that as "prep" and call it one thing?

If the evidence shows that none of that works, then yes it can all be lumped together

That's funny cliff. By the same logic we all have IQ 100, done deal.

Uh, no

I dream of finding employees that used the book vs the course. It's a great measure of wisdom and work ethic.

So what we observe here is intergenerational transmission of study habits.

Perhaps we need to focus on ways to encourage parents with less means to be able to more effectively support their children's education and study habits.

Probably promotion of high quality public education wouldn't work though.

Grasping so hard to any possibility that suggests intelligence is not heritable. You are wrong.

I missed this gem: "Probably promotion of high quality public education wouldn’t work though." Yes, Nate, if only we gave the teachers' union more money, then this whole problem would be solved. Or, maybe we could skip the whole handouts to middle-class white women, and go straight to the cause: poor behavior by poor people. Nah, that would never work, let's just increase handouts to Democrat voters instead.

Some people are smarter than others. Our economy rewards intelligence. Intelligence is partly heritable (with wide variability, but we're looking at aggregate data).

This is another "partial" explanation.

So weird... it's almost as if smarter parents tend to make more money and tend to have smarter children...

LOL. +1

It would be silly to think that is not part of the story.

But is it 1% of the story and 99% explained by social factors? 10% of the story?

My guess is that it's fairly insignificant compared to socially, economically and politically entrenched factors, for example where connected people make more money and their children inherit their networks in a sense.

How can you compare income across generations without considering that some people plain and simple have a head start via opportunity, relatively speaking.

If you take the time to carefully read the most important papers in behavioural genetics you will soon discover it is not 1% or 10% but more like >65%. However this link between SAT type scores and parental SES is stronger in America than in many other western countries. The causes for this are complex, significant factors include the the type and scale of immigration, that SAT tests are closer to IQ tests than similar schemes, racial gaps, but also the stratification of wealth, variable school quality, the history of race etc. We could do better but we must also be honest that there are real limits to how much better we could be.

My parents worked in inner city schools. Dinner table stories were about smart kids who faced daily disincentives to learning. This included ridicule from parents who thought education was a sucker's game. Contrast this to the typical upper middle class home filled with educational toys. Or at the other extreme, the "B+ no good, do it again" Asian home.

65%, not 100%.

I do not think that head start is such an advantage in the long run. It cuts both ways, namely as a de-motivation (why should I work hard when I can simply party for my dad's money?) The Marines say that "complacency kills", and it does.

Sorry, I think this was really a terrible piece.

(1) Just because the relationship is consistent and monotonic does not mean it is "strong." The relevant question would be what is the R-squared? The article doesn't give this information. In actual fact, most of the variation in SAT score is *not* predicted by parental income, so to suggest it's mainly and "affluence test" is grossly misleading.

(2) The purpose of the SAT is not to "reduce the economic gaps in the test." It's to correctly identify those who have the preparation and aptitude to do well at college, whatever their income. Is there any evidence the planned changes will improve the performance in terms of measuring what we care about? (Answer: no)

(3) It trots out the old "regatta" example that hasn't been on the test for decades. If this type of thing was so important, why can't the critics come up with a remotely recent example? And why are the gaps just as large in math as in reading?

Because a teenager will only really apply himself to trigonometry if he needs to navigate his yacht in a regatta.

The issue should be whether SAT test scores correlate with outcomes...a job later, graduating from college, etc. Now it just correlates with admission decisions.

Uh, what? Aren't SAT scores the best predictors of college success (versus grades etc.)?

This is why college admissions should be based on value-added modelling, not unlike traditional value-added modelling on schools and classrooms, or, as of recently, attending physicians.

A study wherein the ideal of meritocracy gets outed as the road to oligarchy

Wait didn't Charles Murray write a book about that. Terribly racist to even bring it up, of course.

On average, students in 2014 in every income bracket outscored students in a lower bracket on every section of the test

That's not very informative because when you bin the income variable into income brackets, that will be true as long as there is a positive and linear correlation between income and SAT scores. No matter how small the correlation, each successive income bracket will necessarily outscore the previous one.

IIRC, the SAT-income correlation is about 0.30. Eliminating all income differences in society would, assuming the relationship is not spurious, eliminate 9 percent of SAT score differences. The SAT can be regarded as just another IQ test (it correlates with standard IQ tests as well as they correlate with each other), so the SAT-income correlation is probably mostly or entirely due to genetic differences.

I don't see anything in that study about controlling for social factors.

Therefore, it is useless.

If you can't see through that, then prehaps you're not as smart as you thought you were, and are only capable of seeing evidence which corroborates an inflated opinion of self worth relative to others?

Nathan W, in a rare moment of self reflection.


Everyone is upset about the "correlation," even though nobody seems to have bothered to figure out how large this correlation actually is, since (as mpowell pointed out above), everyone completely ignores the substantial variation within income groups.

I wonder how the strength of this correlation compares with the strength of the correlation of education levels to income? I also wonder what is the likelihood of people with higher levels of education marrying each other and reproducing, versus marrying and reproducing with someone from the lower levels?

This has been shown many times.

Also, people with similar taste in art, etc. are more likely to get together.

What's your real question?

Read Alex on this!

Another data point to affirm my bias that suggests no force in the universe - no policy, no teacher, no market for education, no funding level, can save you from peer group and parental influences. Kids almost can't be saved from critically timed don't give a crap attitude, and the strongest influences at that critical time is how much give a crap their friends have and to a lesser extent their parents have. Ability yes, but my ex rectum sense is give a crap dominates raw IQ in terms of middle to mid-high range outcomes.

Parents cause the genetics, you can't separate parental influence from genetics, unless you look at adoption studies. And the studies show that adopted children will have outcomes closer to their genetic parents than their adopted parents. +1 genetic determinism

I posit people choose their peer groups, based on genetic factors (intelligence, personality), that is why they appear to be influenced by peer groups. You can not prove otherwise. I've not heard of studies which attempt to artificially change children's peer group. I suspect, from personal experience and observation, that the child would still gravitate to the same type of people who they always gravitated towards.

While the adoption design is powerful for detecting shared environmental effects, adoptees are relatively rare. The main way of disentangling genetic and non-genetic family effects is the classic twin design where MZ and DZ twins are compared. This relies on the assumption that MZ and DZ pairs have equally similar environments, which appears to be true in most cases, or violated to such a small extent that it doesn't matter. In general, it's clear that the family environment cannot be very important because the majority of population variation for most behavioral traits exists within families (i.e., normal siblings are more different than similar to each other).

"normal siblings are more different than similar to each other"

That's madness and certainly untrue. Siblings are more alike than 2 randomly chosen people. Your whole post is deranged, adoptees are not rare vs twins, look at the us statistics and the population is roughly equal annually. I'll stop there....

I don't think you quite understand the issues involved.

When I say that siblings (except for MZ twins) are more different than similar, I mean that their correlation for behavioral traits is generally less than 0.5, which means that more than 50 percent of population variation exists within families. Here's a classic paper on this: The reason for this is that siblings share about 50 percent of their segregating genes, and the shared family environment generally has little to no effect on individual differences after childhood, which means that genetic similarity (0.5) is usually the upper limit for behavioral similarity for siblings.

Adoptees used to be much more common but easily available contraception and abortion have decreased their numbers in the developed world. There are substantial numbers of international adoptees, but they are problematic from a sampling perspective.

I apologize, excellent clarification.

I guess my larger point is there are some environments where pro educational peer influences are much more likely than others. Kids will choose peers, but they are fishing in different ponds when they do so. I recognize the chicken egg issues here. I basically feel like a kid in a school with dominant culture that is anti educational has to be pretty extraordinary in ability and self motivation to get away from that. If the average kid you encounter at a school at least feels education matters, that's a different situation altogether.

How many kids in the "hood" can see evidence around them in their daily lives that education can contribute to earning potential?

I agree with that. I offer no judgement on the kids, I just don't know how to help people who can't leave the environment. I don't think redistribution addresses this in any real way. I support many anti poverty measures. I support public provision for education. All I know is that there is a huge difference between caring and not caring in the attainment of skills and knowledge and that peer groups are profoundly influential. If I were a parent, I think my first plan of attack would be to put my kid in the environment with the most successful kids and I would try like hell to do it by middle school.

I bet "the hood" explains why the 140K-160K group is underperforming the 160K-200K group. All those 140K households in the middle of gangland, it's no wonder they cannot keep up with the 160K crowd.

Dan Weber's comment made me laugh out loud at work.

"I bet “the hood” explains why the 140K-160K group is underperforming the 160K-200K group. All those 140K households in the middle of gangland, it’s no wonder they cannot keep up with the 160K crowd."

LOL. Good counter point.

But what about that study about poor families that became well-off and the new kids they had did not do better? That was a striking finding for me because you would absolutely think the ghetto would have an effect.

We should disregard school quality, the effect of broken homes (perhaps because dad lost his job then got thrown in jail for selling dime bags to pay the bills) on both income and education results, and most especially ignore the presence of opportunities within extended peer and family networks as explanations for why these students obtain lower results on the test.

Accounting for these factors will lead to inconvenient truths which support redistributional mechanisms geared towards equality of opportunity for each cohort of students. Thus, they are to be swept under the rug if and when possible, because everyone knows that competition is bad for progress, and the people with a leg up can't take the heat of equal opportunity.

We should also ignore peer effects. Surely they couldn't multiply the above factors .

Okay, but what is the effect of those things on SAT scores?

Perhaps what we observe is an intergenerational transmission of materialist obsessions and drives where essentially unhappy and emotionally insecure people can point to indicators of their success, such as high test scores (of their children) and fat bank accounts, whereas other groups may pass on more meaningful knowledge of how to make the experience in our blink of the eye worthwhile?

You can almost hear the gurgle of the bong water in that comment.


If you really think that's the case, then surely there is no point fretting over inequality or disparities in outcomes re: test scores, then right? Because there are a lot more "meaningful" things to do than score highly on some standardized test or accumulate a lot of cash, right?

But actually, that completely cuts against the content of your previous comments, so I suspect you don't actually believe that.

Nathan, this is just a step away from the Index of National Happiness, as introduced by the King of Bhutan. He needed at least one variable for his country to excel at.

More seriously - during my life, I encountered lots of people from various income and wealth backgrounds, and my own income was oscillating over the years quite significantly, so I do speak from some experience. It was the Czech Republic, and not the USA, but people are similar across the West.

I can't say that poorer people are less materialist or have more "meaningful knowledge" than the richer ones. In fact, the opposite seems to be true, at least when it comes to spending for unnecessary stuff. Plenty of poorer people I know got into dire straits precisely because they bought a lot of shiny stuff and got into debt.

He's a reporter, not a social researcher. It has been remarked that Journal reporters (in contrast to editorial writers) have citation patterns in public discussion which resemble those of Democratic members of Congress).

It hardly seems to occur to him to ponder the direction of causation, or consider that perhaps the phenomena under discussion might both be generated by an unobserved factor. Hack.

Most people, including most reporters, aren't very sophisticated consumers of statistics.

I've heard the same is true of statisticians. Of course, I don't have data to back that up. ;-)

Tangential, but for your reading pleasure. At an elite law school, a hallway debate took place between a Jewish student and a Muslim student, over predictable political topics. The argument reduced the Muslim girl to hysterical tears, at which point my unlucky roommate rounded the corner.

Joe was an Italian kid raised in Queens, first by an abusive, alcoholic father and later by his cash-strapped single mother.

Fast forward to the hallway, where Joe rounds the corner and stops short with a stunned expression at the scene before him. A friend of the crying girl wheels on Joe and says, "I hope you're happy, you privileged white f---."

Quiet You, don't you know the struggle that Sasha Obama will go through in her life?

Thats how it should be, otherwise that money for private school goes to waste. $=Results, always has always will. it doesnt mean no $ = no success, but $ always gets results

I wish

It is true within country but not true between countries, e.g.
Harvard historian Niall Ferguson on breakdown of OECD PISA score with respect to parents' occupations,

Are we allowed to ponder the thoroughly replicated fact that the children of the lowest decile of low-income whites outperform the children of the highest decile of high-income blacks on these tests?

I believe in the most recent numbers that is no longer true. Doesn't change the point though.

Here's the data from 2008:

Actually, I wrote about test prep and race a good year before Alex did, and it still annoys me he somehow found all the relatively obscure links from my piece and yet never mentioned my article.

Of course, this whole topic is absurd because race is a much better predictor of SAT scores than income is. Poor whites outscore high income blacks on the SAT (as well as on every other metric).

It is interesting that the top income group only averages below 600 on the math part of the SAT.

Quality content is the key to be a focus for the visitors
to pay a visit the site, that's what this web site is providing.

Comments for this post are closed