Coding Error

James Coleman published an influential study in 1966, “Equality of Educational Opportunity”. It found that variation in school quality (per pupil expenditure, size of school library, and so on)  had little influence on  educational achievement, while students’ family backgrounds did.  Which is consistent with genetic influences.

He did find that a student’s educational attainment was related to that of other students in the school – somehow, educational success rubbed off on others. This was often cited as evidence to support policies of forced integration and busing.

But that finding was a mistake – due to a coding error.  Which I first thought meant missing a parenthesis in some computer program, but more likely had to do with mislabeling or misclassifying entered data.






This entry was posted in Uncategorized. Bookmark the permalink.

71 Responses to Coding Error

  1. Mark R. says:

    The ultimate error here to the proximate one of the coding error is the influence of Justice Louis Brandeis and the “Brandeis Brief”. That’s how “science” came to be used by progressives to override legal citations and principles and to promote large social changes through the judiciary:

    “The Brandeis Brief changed the direction of the Supreme Court and of U.S. law. It is considered a model for future Supreme Court presentations in cases affecting the health or welfare of classes of individuals. This strategy of combining legal argument with scientific evidence was later successfully used in Brown v. Board of Education to demonstrate the harmful psychological effects of segregated education on African-American children.[10][11]”

    • david says:

      It would be interesting to make a documentary on how wrong the application of that “evidence” was. Im pretty sure african americans have worse avg math and reading scores now than those of 60 years ago.

      • cameron232 says:

        I keep seeing the claim that kids nowdays are way dumber, that a college freshman of today couldn’t pass an 8th grade test from 60 years ago or whatever.

        Both my uncles (lived in a medium sized town) were considered smart because they took Trigonometry in their senior year of high school.

        • Regret says:

          60 years ago the goal of schooling was that stdents should be taught. Now the goal is that no child should be left behind. If you want everyone to graduate, graduation has to be something everyone is capable of.

          But that doesn’t mean the kids are dumber

          • gcochran9 says:

            They are dumber, mostly due to demographic change.

            • James says:

              Are you aware of any study that tries to crunch the numbers to see if the Flynn effect was outweighing demographic change until recently? My impression (if we are talking about 1960-2020, 60 years) was that the Flynn effect definitely had a positive impact on US IQ scores from 1960-1990 before stalling (probably due to nutrition + environmental quality finally allowing close enough to everyone to reach their potential), and that that could possibly outweigh demographic changes.

              I’m not saying that it has, just that I would have thought you’d need to run the math to see and I don’t know where to get trustworthy Flynn effect / demographic IQ data (if it exists in a time series form).

            • Regret says:

              The premise was comparing African-Americans of the past and p

            • Kilo 4/11 says:

              I would be grateful if you’d elaborate a bit on “demographic change”. In my limited understanding, demographics deals with the composition of a given physical area, such as a city, zipcode, county, etc. in terms of the balance of race, sex, income, educational achievement, etc. to be found therein. I can’t seem to connect the dots between these categories and kids being dumber today, though I do believe they are.

              • James says:

                He’s referring to the fact that at a macro level it is well-documented that average white IQ [~100], average Asian IQ [~104], average Hispanic IQ [~90], and average Black IQ [~85] differ, which means that the average IQ of the average American child has probably gone down {it is disputed how much of Hispanic / Black IQ being lower is environmental vs societal vs genetic, and how much of Asian IQ being higher is genetic vs immigrant selection effect}.

                But, as I mention above, it’s only probably because at the same time the Flynn effect has been resulting in the 100 of white average IQ to be moved up –> IQ scores are normalized such that white IQ is always average 100, and the Flynn effect [mostly positive environmental factors like nutrition] means that “100” in 2020 is higher than “100” in 1960.

                I was hoping that someone had crunched the numbers comparing these factors to see which one predominated.

            • X says:

              Which factor is greater here: Dysgenics or immigration?

        • DRA says:

          I am always amused by newspaper articles that talk about how little high school graduates know about geography and (recent) history. I have the advantage of having read about both topics over the last 60 years, most of it having changed or happened after graduated from high school.

        • Mark R. says:

          Trigonometry was the highest level of math taught at Richard Feynman’s high school when he went there in the ’30s.

          • Rosenmop says:

            Algebra and trigonometry was the highest level of math taught in grade 12 in the early ’70’s in British Columbia, Canada.

      • pyrrhus says:

        The tests have been dumbed down considerably in the interim….

  2. Space Ghost says:

    With the rise of unsupervised learning, we’ll be forced to manually implement coding errors of this sort, in order to prevent the machines from being too good at recognizing reality.

  3. teageegeepea says:

    What made you change your mind?

    • gcochran9 says:

      I ran into various sociological papers in which they’d have some minor problem, like getting all the results backward, and the explanation often involved something like casually reversing the code numbers for two groups, or something similarly stupid. They called this ‘coding errors’.

  4. morris39 says:

    Coding error! Is that assertin based on some evidence? If not easily demonstrated, then is that statement a good indicator of viewers’ mood affiliation ?

    • gcochran9 says:

      The error was noticed after a bit: Coleman talked about it, I believe. As did others.

    • Gringo says:

      Coding error: Coleman Report

      Remarkably, considering both the constraints of time under which he and his colleagues worked and the limited agenda set by the government bureaucrats who monitored the study, all but one of the major findings generated by Coleman withstood subsequent examination by an army of social scientists. (For example, a group of eminent social scientists and social statisticians formed the Harvard University Faculty Seminar on the Coleman Report, and met regularly for a whole year, for the sole purpose of verifying the original findings by reanalysing the data.) Subsequent reanalysis showed that a coding error had produced greater evidence of peer effects in schools than was actually the case, a particularly unfortunate mistake, since this finding was often cited as evidence to support policies of forced integration and busing as the most effective way of ending racial segregation and raising Black educational achievement.

  5. skeptic16 says:

    I think in this case, “coding” refers to how data is classified for analysis rather than a computer software error.

    • bob sykes says:

      In the 60’s and 70’s, that’s all “coding” meant. “Programming” referred to the writing of the instructions. Apparently, nowadays “coding” is a synonym for what we back then called “programming.”

      • skeptic16 says:

        Coding is what journalist suggest as a new career for those that lose their jobs but react angrily when the suggestion is made to them after THEY get laid off.

      • dearieme says:

        That’s spot on, bob. In the 60s I “programmed” my mathematical models, I did not “code” them.

        I wonder why the word changed? Easier to spell?

  6. William O. B'Livion says:

    I suspect there is at least a small relationship between the education success of some students and others in their class/school.

    Genetics (roughly) defines our maximum abilities–how fast the nerve can twitch, just how smart someone can be.

    But it doesn’t mean that you will necessarily get to that potential, just that you can given the right conditions. Most kids might be a little smarter if they had more EPA/DHA in their diet (and their mothers diet in utero).

    But certainly when children aren’t expected to learn, and teachers don’t encourage them to push themselves, and (at least for boys) don’t have the competition to be better at school, then they won’t.

    In schools where where all children are expected to do better, almost all children will.

    • gcochran9 says:

      “almost all children will.” Silly.

      • James says:

        Isn’t this the pattern we see for “No-Excuses” charter schools?

        Schooling can’t raise IQ, but it can ensure that children actually get to use that IQ in more productive ways. The success of “No-Excuses” extremely strict charter schools in urban environments seems to fit a model of “lower average IQ population needs stricter schooling to learn, but can in a stricter environment achieve more educational attainment than in a less strict environment”** and also “removing children from pathological subcultures that devalue education allow them to get closer to their maximum possible educational attainment” -> unclear which of these two factors is more significant, but both seem very likely to have an impact.

        Notably, white progressives hate “No-Excuses” charter schools, even though they (1) are majority black/brown students, (2) appear to be benefiting those students, over and above a selection effect for brighter students, (3) are voluntarily opted into by parents.

        White progressives probably hate these schools for some combination of (1) distaste for strict education, since a strict schooling model appears to be a throwback to a more conservative, less “enlightened” time (2) charter schools drawing resources after from public-sector teachers unions, which are an important part of the progressive coalition (3) [maximally cynically and which I don’t believe] these schools helping black/brown children succeed means more competition for the children of white progressives.

        **Stricter schooling benefiting a lower IQ population would also apply to the entirety of the US in general a century ago, when white IQ was far lower due to malnutrition and iodine deficiency and other factors. Farmboys from interior states that were drawn from an iodine-deficient IQ85-average population still got through schooling with useful skills and the ability to be assets to the WWI US army.

    • Hugh Mann says:

      “But certainly when children aren’t expected to learn, and teachers don’t encourage them to push themselves, and (at least for boys) don’t have the competition to be better at school, then they won’t.”

      It does make a big difference what the cool kids do at school. Whether they’re skipping out to hit the betting shop or arcades after lunchtime or they’re getting good grades.

      There’s a reason why parents have attempted to control their kids peer groups since time immemorial.

    • Frau Katze says:

      Bad peers can affect a child, but it wears off as they become adults. That’s just what I’ve observed.

    • Anonymous says:

      “almost all children will” – not really. That is, there is 10% of IQ variance explained by non-shared environment (i.e. things outside home environment, like school), but that’s it. 60% to 80% of IQ variance is explained by heredity. See

  7. ghazisiz says:

    The view of “peer effects” that led to “integration”, “forced busing”, and “mainstreaming” was that peer effects were asymmetric: bad students would become more like good students, and good students would not become worse. In fact, as anyone with any experience of teaching could have predicted, bad students grabbed most of the teacher’s attention, and good students were neglected — an outcome that colleges of education still won’t openly acknowledge (coz raciss).

    • saintonge235 says:

      And bad students disrupted the classroom, affecting teaching for the worse regardless of the where the teacher’s attention was. I noticed and complained about this when I was in high school, 1967-71.

      • Henry Scrope says:

        My wife tells me, from her personal experience, that disruptive pupils are now almost impossible to control, discipline, suspend or expel, to the detriment of the majority.

        She says the ‘rights of the child’ are exclusively the right of the problem child.

        • another fred says:

          That was my experience when I tried to “give back” by doing a little teaching in retirement (I’m an engineer and though my grasp of real world application in STEM would be of value). The lunatics are running the asylum.

          The school administration has a long drawn out procedure for dealing with the disruptive students that assures loss of control of the classroom. The “Overton Window” for tolerated behavior is skewed heavily towards chaos. I feel for those who are trapped in teaching as a career. They see the reality but are pretty much helpless to do anything about it.

          • saintonge235 says:

            Nowadays, the great concern is disrupting the “school to prison pipeline.” IOW, not sending criminal teens to jail.

            It isn’t only that the lunatics are running the asylum. It’s that the asylum is being sabotaged from within by those opposed to its mission.

  8. James B. Shearer says:

    I thought that a peer effect meaning that everybody is pulled a little towards the mean was fairly well established. Is that not the case?

  9. oldmiseryguts says:

    Just testing to see if I’ve been banned/blocked.

    • Regret says:

      60 years ago the goal of schooling was that stdents should be taught. Now the goal is that no child should be left behind. If you want everyone to graduate, graduation has to be something everyone is capable of.

      But that doesn’t mean the kids are dumber.

  10. oldmiseryguts says:

    Apparently not.

  11. Lior says:

    Bill gates small classroom initiative was based on the fact that the smallest classroom had the highest scores,of course they also had lowest scores as a small sample size means more variance.It only costs gates several hundred millions to learn that, there by proofing the high costs of education.

  12. linsee says:

    Coleman also admitted, in a speech he gave at the National Association of Scholars convention in 1990, that he didn’t follow up on the observation black children did worse in schools with many black teachers. I have a copy of the speech, which was published in Academic Questions – Greg, would you like me to send you a copy? – and I wrote about it last year”

    “Coleman also said that black children did better if they attended schools that
    were predominantly middle-class (but only if most of the other students were white).
    “That observation (and subsequent lawsuits) led many school districts to
    adopt policies placing black children in mostly-white classrooms, which, in turn,
    led many parents to leave the districts that had adopted such policies. It devastated
    many urban school districts.

    “But looking back much later, in a speech he gave in 1990 (I heard it) Coleman
    acknowledged that there was one question he did not ask at the time, a question
    whose answer might have been as disruptive as busing for racial balance. Teachers
    in the survey took a vocabulary test, and scores on that test were associated
    with children’s verbal achievement. Black teachers in segregated schools, themselves
    often educated in segregated schools, ‘were on the whole less prepared
    and qualified, with lower verbal skills than their white counterparts.’
    “The question Coleman and his colleagues did not ask, and didn’t want to ask,
    was: How is the achievement of black children related to the racial composition
    of the teaching staff of the school?”

  13. Regret says:

    The premise was comparing African-Americans of the past and present. I dont think they’ve lost ground. I don’t expect any specific racial group will have.

    The demographic balance is worse, yes.

  14. Gringo says:

    More documentation on coding error: Jon Clark (Ed.): JAMES S. COLEMAN (CONSENSUS AND CONTROVERSY FALMER SOCIOLOGY SERIES) page 93

    However, Coleman’s empirical claim about the importance of peers in schools for the educational performance of children did not hold up in subsequent reanalyses of the Coleman data (Smith, 1972; Hanushek, 1968, 1972). As noted by Smith (1972), a simple coding error produced much greater evidence of peer effects than was found in analyses of the corrected data. Unfortunately, this evidence on peer effects had considerable impact on policy and served as an
    important catalyst for busing and forced integration programs. Since home environments are less easily influenced by policy, Coleman’s evidence was cited as evidence that forced integration was the most expeditious policy for raising black achievement.

    p 474 Smith, M. 1972. ‘Equality of educational opportunity: the basic findings reconsidered’, in F. Mosteller and D.Moynihan (eds), On the Equality of Educational Opportunity. New York, Vintage Books.

  15. Gene Environment Correlation is a challenge here. A couple potentially relevant points:
    – Bouchard’s MZAs were significantly different from Bouchard’s MZT’s. Even the MZAs had correlated environments to a significant extent.
    – The adoption studies show less c^2 than Bouchard’s twins. I am not saying the adoption data or twin data contradict each other, just that the differences require explanation.

    Inter-generational GWAS should help a lot here, but I’m curious how you reconcile the above without a small but significant level of c2. Also unclear to me how such confounds work with peer effects in communities with pop strat.

  16. skeptic16 says:

    Prince Georges County MD is 2/3 black and has a median household income of $82,000 which is well above the white median household income of the US. Yet, 68% of their public schools are considered below average by greatschools. Clearly, there is more than income or whether the parents have stable employment that determines educational outcomes.

    • X says:

      A lot of blacks there work for the US government, no? Possibly as a result of the US government engaging in mass affirmative action hiring of blacks.

  17. Gordon William Marsden says:

    i once had a university math test that was overly difficult , 4th year of engineering so difficult to judge us all as study, the solution was to give everyone a mark that was 10 x the square root of your test score. so 100 received 100 ( none did) 50 got 70 , yet 0 still got zero., easily coded or miscode and generated a believable curve . they published the curve at the top of the listing all the time for some reason. lost in the cobwebs of my mind what i attained but i did pass

  18. Question says:

    Did Coleman ever try raising dachshunds and greyhounds in the same pasture, with the expectation that the weiner dogs would learn to run very fast?

    • Kilo 4/11 says:

      The weiners won’t be running any faster but the greyhounds will become foul-mouthed druggies and wear loud clothes. They’ll also get the message that running fast is a bad thing and start getting slower.

  19. X says:

    I have a question for Greg Cochran: What are your own thoughts on forced integration and forced busing?

  20. David Chamberlin says:

    Hi Greg. I am reading another big history book called Sapiens, A Brief History of Humankind by Yuval Harari. It’s garbage that needs to be countered. I think I will be able to read it all but I might throw it across the room a few times. It’s as if a new generation of ignorant Jared Diamonds have sprung forth from the belly of the beast. This yahoo recycles all the usual nonsense that evolution stopped cold way back yonder back when the first lion man was carved in a cave during the ice age. Please bludgeon this fool with the overwhelming evidence, I know it won’t do any good but someone has to try.

    I get why they have to play make believe that the human brain is exempt from evolution. Once you start down that road it opens a Pandora’s Box. If evolution is ongoing then maybe there are shifted bell shaped curves of intelligence between populations but right about then all hell breaks lose and rational thought is gone in a puff of smoke. Honest men can’t get their books published, they lose their university jobs, they are branded as hate filled racists. Lot’s of unwanted truth about better brains.

    What are going to do.

    • j says:

      Yuval and his husband are busy preparing the comics version of the book. The overwhelming evidence you are suggesting should also be in comics. To be even more convincing, the Lascaux cave paintings may be included (with balloons).

      • David Chamberlin says:

        in fairness to Yuval Harari the book gets better after the first few chapters, he doesn’t know what he is talking about regarding recent human evolution or human prehistory in general. But once he gets nearer to the present he has some thought provoking ideas. I just wish there were a few more big history books written where the author would research his subject areas better and clearly separate interesting and substantiated facts from wild speculation.

        • ghazisiz says:

          Students have approached me after class and asked if I had read “Sapiens” — they had read it and found that my lecture resonated with what they had read. After all these years of teaching, I have decided that that is just the way the world works: many voices, none of them (including my own) really knows what has happened and what is most probable (often the difference between 0.99 and 0.98). A student can only take the kind of path that I took as a student, listening to everybody and gradually deciding as they read and think things over. If there is a problem with this process, it is that the current generation seems so different from their predecessors: they focus their eyeballs on social media, and they do not find any joy in opening a book and losing themselve in it.

          • David Chamberlin says:

            Wise thoughts. It is easier to maintain one’s ignorant beliefs today, a whole channel of talking heads will confirm whatever shit it is you want to believe, but it’s also easier to find really good non fiction.

  21. General Fairfax says:

    Greg, can you please do an update on Richard Epstein’s covid-19 predictions from earlier this year?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s