The form is one side of a sheet, printed cheaply, and I have a copy of it from a low-fee private school in a district town where I ran a workshop for two years. Name of student. Date of birth. Previous school. Mother tongue. Father's occupation. Mother's occupation. Annual family income, with four ticked ranges.
Nothing on that form is illegal and most of it is required by somebody. The income bands are there because of a fee-concession quota. The mother tongue is there because of a language-policy circular. The occupations are there because they have always been there.
The school has three sections per grade. Parents refer to them, without embarrassment, as A, B and C, and they will tell you which is which.
I have been trying for six years to establish, in a form I could publish, what determines the section. What I can say with confidence is what everybody in the town says, which is that the form determines it, and that the occupation columns matter more than the income bands, and that this is understood by every parent filling it in and denied by every school I have asked.
Let me stay with the form for a while, because the mechanism is in its details and not in the general point.
The occupation columns are free text. A parent writes what they write. What arrives is a mixture of job titles, employers, and — this is the part that took me longest to notice — self-descriptions that encode a great deal more than a job. Government service. Own business. Agriculture. Private job. Household work. Each of those is doing something different. Government service is a status claim and everyone reads it as one, because it means a salary, a pension and a certain kind of family. Own business covers a man with three employees and a man with a handcart, and the school knows which is which because the school is in the town. Agriculture covers eleven acres and half an acre.
So the column does not collect an occupation. It collects a signal that a local reader can decode with high accuracy and that a researcher forty kilometres away cannot decode at all, which is precisely why the practice is invisible in any dataset.
Then there is the mother's column, and this is where I stopped being able to treat it as an administrative artefact. In the copies I saw, the mother's occupation was filled in for a minority of forms. Where it was filled in, the entries divide almost cleanly into two kinds: professional titles, and the word household. Where it was left blank, the household reading is also available. What that column tells a school is not what the mother does. It tells the school whether this is a family where a woman having a job is something to write down.
And the income bands, which are the only part of the form designed to sort by economic status, are the least informative thing on it. They are self-reported, unverified, and everybody rounds toward the concession threshold. A school that wanted an economic sort would ignore them and read the occupations, which is what I am arguing it does.
Now the school's answer, which I have been given three times in almost identical words and which is not a stupid answer.
Sections are grouped by preparedness, not by parentage. The children arriving from an English-medium pre-primary can follow instruction from week one and the children arriving from an anganwadi or from home cannot, and putting them in one room means teaching to neither. The grouping is pedagogical, it is revisited after the first term, and the correlation with the father's occupation exists because prior schooling correlates with the father's occupation — which is a fact about society, not a decision by the school.
Every clause of that is true and the last one is the one that does the work. Yes: the correlation is upstream. But a school that sorts on prior schooling, at age five, into sections that are then taught differently, has taken a correlation and converted it into a structure, and the structure has its own effects for the next eight years. The teachers assigned to C are not the same teachers. The pace is not the same pace. A child does not move from C to A after the first term; in six years of asking I have found two cases and both were siblings of an A child.
The mechanism does not require anybody to intend anything. That is what makes it durable and what makes it hard to write about, because there is no villain and there is no memo, and everybody involved can describe their own decision as reasonable and be right.
There is one more thing the form does and it is the reason I keep the copy.
A family that understands the form can play it. I know of a household — a carpenter with a small workshop — where the father wrote own business and the mother, who does tailoring at home for money, wrote nothing at all, and the child went into B. The parents had discussed it. They told me they had discussed it. They were not confused about what the form was for; they were completing it strategically, with a fair understanding of the decoding, and they got a better outcome than the form would have given them.
Which means the sorting is not even a clean sort on the underlying thing. It is a sort on the underlying thing plus a family's literacy about the system, and that second term is itself distributed exactly as you would expect.
My proposal is the smallest one I have: take the occupation columns off the form. Not the income bands, which are required for the concession quota and are at least honest about what they are for. The two free-text columns, which no circular requires, which no school can explain a use for when asked directly, and which are the only fields on the page that a school could not defend in writing.
I asked a principal what would break if they came off. He was quiet, and then said he supposed they would simply have to ask the parents in the interview instead, and then he heard himself say it, and neither of us said anything for a moment.
