Think of the one or two children in your primary school class. Not the ones with the best marks — the ones who talked to teachers the way you talk to an adult: who would ask "could I do this assignment a different way", who would put in a word for themselves when told off. Everyone else just nodded. The thing looks too small to mention, and the return on it is absurd.
This page takes apart three things: school says it is picking talent, so what is it actually measuring; how the set of lessons that is not on the timetable gets learnt; and why, as everyone studies longer and longer, the same certificate buys less and less. The strongest version of each account goes on the table, with a label for how hard the evidence is; the judgement is yours.
The American sociologist Annette Lareau and her assistants got that small thing on record. In the mid-1990s they picked twelve households out of an interview sample of 88, and spent three weeks with each around a child of about ten: at the dinner table, at the edge of the ball field, in the doctor's office, at parents' evening.
She saw two logics of raising children. Middle-class families packed the child's schedule with organised activities, reasoned with them, and encouraged them to make requests of teachers and doctors (she calls this concerted cultivation). Working-class and poor families got food, clothing and safety solidly handled and let the child grow on their own (the accomplishment of natural growth), and dealt with institutions more by keeping a distance and not causing trouble.
The point is not which one is better. It is that one of them happens to match what institutions like schools and hospitals expect. In the surgery, one child interrupted the doctor with a question and the doctor explained two things more; another had been told beforehand not to talk so much, and came out with nothing extra. The same behaviour is called quick in one setting and rude in another — and the ruling is made by the institution, not at home.
The harshest statement of this chain comes from Bourdieu and Passeron. In 1964 they set out the social origins of French university students: in the academic year 1961–62, children of senior executives and the intellectual professions had roughly 42 times the chance of entering university that children of workers had.
Their real target was not that gap but how school makes it look deserved. Everyone sits the same exam and is ranked by the same ruler, so those who lose charge the result to themselves — not clever enough, not hard-working enough. School therefore does not merely pass the gap on; it translates origin into talent, and talent into merit. That step is what they call méconnaissance: the real mechanism is taken by the participants for something else, and the people who get it wrong have signed off on it themselves. As for what the ruler measures — how you speak, whether you stay at ease in front of authority — the page on cultural capital and habitus covered how those get accumulated at home, day by day.
Two criticisms of this account are hard ones. First, it is difficult to falsify: any outcome at all can be described as reproduction. Second, in later quantitative work, cultural capital does have measurable effects on attainment, but they are generally smaller than the original theory expected, and they vary a lot between countries. And one fact cannot be skipped: educational expansion really has put more people through school — while, as the page on social stratification separated out, absolute mobility and relative mobility are two different numbers. What holds up is that chain of translation, not "school does nothing". Evidence: Lareau is deep fieldwork on twelve households, giving mechanism rather than distribution; the French figures are administrative statistics from over sixty years ago.
Bells, raised hands, queues, handing work in on time, dropping one task for another on command. Not one of these is written on the timetable, and each is drilled dozens of times a day. Drilled long enough, it stops looking like a rule and starts looking like the way the world is.
In the 1978–79 school year the education scholar Jean Anyon spent a year in each of five New Jersey elementary schools, watching what "doing schoolwork" actually looked like in fifth grade. She sorted the five by the occupational make-up of their communities into four tiers: two working-class schools, one middle-class, one in an affluent professional community, one executive elite.
The same arithmetic, four versions of what counts as doing it well. In the working-class schools, work meant following the steps, and getting it right meant having followed them. In the middle-class school, work meant arriving at the correct answer, by whatever route you chose. The affluent professional school stressed expression and creativity: what you handed in had to carry your own ideas. The executive elite school stressed analysis and reasoning, and students were required to answer for their own output. What is taught daily is not the content but which kind of authority relation makes work count as done right. This unwritten syllabus has a name: the hidden curriculum.
Bowles and Gintis gave a harder version in 1976, the correspondence principle: the organisational form of school (grading, punctuality, obeying superiors, working for external rewards) corresponds one-to-one with the organisational form of the workplace. The criticism it draws is that this is too mechanical, treating students as blank pages to be written on.
The strongest rebuttal comes from a piece of fieldwork (covered on the page about Marx's sociological legacy): in a boys' school in the English Midlands in the early 1970s, Paul Willis followed twelve working-class boys for a year and a half, then followed them into work for another six months. These lads simply did not believe that studying hard would get you up; they mocked the compliant openly and made manual work into a matter of masculinity. And it was precisely that resistance that walked them into the shop floor themselves. Reproduction does not require anyone to believe in it; seeing through it and doing the opposite can land you in the same place.
One classic study here needs its current status stated. In 1968 Rosenthal and Jacobson's Pygmalion in the Classroom picked students at random, told teachers they were about to bloom, and reported at year's end that this group had gained on intelligence tests. Three decades of replication went badly: Raudenbush's 1984 meta-analysis pooled a very small effect (d ≈ 0.14), and where teachers had already known the students for more than two weeks before the expectation was planted, the effect all but vanished. What stands is much narrower: teachers do treat different students differently, but "expectations can substantially raise intelligence" has not replicated.
Evidence: Anyon and Willis are both small-sample ethnographies, giving mechanism rather than distribution; the expectancy result is a meta-analysis with a small effect under tight conditions.
Degree requirements in job adverts keep being raised while the work barely changes. This is not an impression; it can be counted.
Take the stock first. The share of people aged 25 and over in the United States holding a bachelor's degree or higher was 4.6% in 1940 and 37.9% in 2021. Same piece of paper, eight times as many holders.
Now the threshold. In 2017 Joseph Fuller and Manjari Raman at Harvard Business School analysed 26 million job postings: for the job of production supervisor, 67% of the adverts demanded a college degree, while only 16% of the people then working as production supervisors had one. The bar sits well above the people currently doing the job, and the job is the same job.
The sociologist Randall Collins named the process in 1979: credential inflation. His argument is that educational expansion is driven mainly by status competition rather than by the skill demands of jobs — a credential behaves like a currency, worth less the more of it is issued, so everyone is forced to study longer and the bar rises to follow.
Economics has a sharper version. Michael Spence's 1973 signalling model says that even if school teaches nothing at all, so long as getting through it costs less for the more able, the credential still sorts people and still commands a wage. It leaves a checkable trace: the jump in earnings in the year a degree is completed is clearly larger than the increment from one more year of study (the sheepskin effect). In 1996 Jaeger and Page recomputed this using survey data that recorded years of schooling and actual degrees separately, and put the bachelor's jump above 20%.
But pure signalling is too strong. Natural experiments built on reforms to compulsory schooling — the law changes, a cohort is forced into an extra year, and they are comparable with those who just missed it — generally show a real causal return to education in earnings. Both components are present; how much each accounts for is still unsettled, and this is the place on the page that most deserves "no answer yet".
What matters more mechanically is the positionality of a credential: part of its value comes from who you are ahead of. Everyone studies three years longer, the ranking is unchanged and the costs all rise; and nobody can withdraw unilaterally — whoever does drops behind at once. This is exactly what follows an expansion: when the bachelor's stops separating people, employers look at the institution; when that stops separating them, at the master's, at the internship. The sorting device moves up a level, and the sorting has not paused for a second. With every level it moves up, more of the cost lands on the family's side — back on the chain from "the answer was right, the way of saying it was wrong".
These three sections support no "therefore we should do X"; that is a value judgement, outside this discipline's measuring range. What can be said is that the three mechanisms have to be aimed at separately. Widening admission moves the positional one; the school's ruler and what the classroom rewards daily do not budge. Their evidence differs too — positionality has large-sample data behind it, the other two only the mechanism fieldwork supplies. Treating them as one thing is the commonest error on this topic.