A girl in Nithari does not drop out of school on a Tuesday. She drops out over about fourteen months, in small increments nobody logs. First she misses Mondays because her mother has a cleaning shift. Then she misses the week her younger brother has fever. Then the maths gets away from her, and missing school stops feeling like a loss. By the time the school register formally marks her absent long enough to matter, the decision was made months earlier, at home, by degrees.
That is the part most people miss when they think about girls' education in urban-poor settlements. The failure is not dramatic. There is no single villain, no one refusal to be argued with. There is a slow accumulation of small frictions, and every one of them has to be removed separately.
We run Project Uday in and around Nithari, Noida, and this is an honest account of what the work actually involves. Not the brochure version. The version with the costs in it.
The legal cliff at Class 8
India's Right of Children to Free and Compulsory Education Act, 2009 guarantees free and compulsory education for children aged six to fourteen. In practice that means the guarantee runs out at Class 8. After that, a family is on its own: fees, uniforms, books, transport, and the opportunity cost of a fourteen-year-old who could instead be earning or minding siblings.
This is why enrolment charts for urban-poor communities look healthy until Class 8 and then fall off a shelf. UDISE+, the government's school-level data system, has consistently recorded higher dropout at the secondary stage than at the elementary stage. The National Education Policy 2020 sets universal access up to Class 12 as a goal, which is the right target, but a policy goal is not a legal entitlement and it is not a bus pass.
In our experience the single highest-leverage intervention is not in Class 8. It is in Class 6 and 7, eighteen months before the cliff, when the family has not yet started thinking of the girl as an economic unit. Once that reframe happens at home, you are arguing against a decision instead of shaping one.
Six barriers, and what actually moves each one
1. The household work burden
The eldest daughter in a household where both parents do daily-wage work is, functionally, the second mother. She cooks, she supervises younger siblings, she is home when the water comes. School attendance competes directly with this, and the household usually wins, because the household's need is today and the school's benefit is in eleven years.
What works is reducing the conflict rather than arguing about priorities. Evening and early-morning learning slots that sit outside the water-supply and cooking windows. Crèche or younger-sibling activity space running at the same hours, so the girl is not choosing between class and childcare. It is unglamorous and it is the difference between 40 percent attendance and 80.
2. Safety in transit
The distance from the settlement to a government senior secondary school is often two to four kilometres. That walk, alone, at fourteen, past construction sites and market lanes, is the reason a large number of families give a hard no. They are not being unreasonable. They are doing a risk calculation with information we do not have.
Escorted group travel works. Fixed pick-up points, a known adult, a roster the mothers can see. It costs real money every single month and it never appears in a donor's mental model of what education funding buys. We would rather say that plainly than pretend the line item does not exist. Families also respond to knowing that every member of staff is trained on child protection law and on their reporting duties under it.
3. Family resistance is usually a proxy
Very few fathers say they oppose education. What they say is that the school is far, that the girl is needed at home, that there is a cost, that a neighbour's daughter had a problem. Treat the stated objection as a proxy and go looking for the real one.
Most of the time the real one is one of three things: money, safety, or reputation within the community. Each has a different answer. Money is answered with a costed plan the family can see. Safety is answered with the escort roster. Reputation is answered by other families, not by us, which is why community mobilisers who live in the settlement outperform outside staff by a wide margin.
You cannot argue a family into keeping a girl in school. You can remove the reasons they have for taking her out, one at a time, and let them make a different decision.
4. The learning gap the pandemic widened
Schools in Uttar Pradesh were closed or partially closed for much of 2020 and 2021, and children in households without a smartphone, without a second smartphone, or without data simply stopped learning for two years. Many of them were then promoted anyway.
The consequence is a child sitting in Class 9 reading at a Class 3 level. The ASER 2023 ‘Beyond Basics’ survey of 14 to 18 year olds found that around a quarter of those surveyed could not fluently read a Class 2 level text in their regional language. A girl in that position is not going to be rescued by attendance alone. She will sit through lessons she cannot follow until she stops coming.
Remedial teaching grouped by actual reading and arithmetic level, not by age or grade, is the only thing we have seen work. It requires assessing every child at intake and again at intervals, and it requires accepting that a fifteen-year-old may need to start at a Class 3 competency. Programmes that skip the assessment because it feels demoralising end up teaching to an average child who does not exist.
5. Documentation and admission
Admission to Class 9 in a government school requires paperwork that migrant and informal-sector families frequently do not have in usable form: birth certificate, Aadhaar with a matching name spelling, transfer certificate from the previous school, proof of address, caste or income certificate where a concession applies.
Section 14 of the RTE Act states that no child shall be denied admission for lack of age proof, and Section 12(1)(c) reserves 25 percent of entry-level seats in private unaided schools for children from weaker sections. Both provisions are real and both are routinely under-used, because the person at the admission desk has not read the Act and the parent has no way to press the point.
A large share of our field team's time goes into this: name-spelling corrections, chasing a transfer certificate from a school that has shut, escorting a mother to a block office. It is the least visible work we do and among the most decisive.
6. Menstruation and infrastructure
A functioning, private, lockable toilet with water is the difference between a girl attending all month and attending three weeks of it. Where the school's facility does not work, absence clusters monthly and looks in the register like general irregularity. Anyone reviewing attendance data without asking about toilets will misdiagnose the cause.
Why “just build a school” is the wrong mental model
Nithari is not short of school buildings. Noida has government schools, aided schools and private schools within reach. The binding constraint is not classroom supply; it is everything that happens between the child's home and that classroom, and everything that happens to her comprehension once she is in it.
Capital is easy to fund and easy to photograph. Recurring costs are neither. The honest cost structure of a programme like this is roughly: staff salaries, escorted transport, remedial teaching materials, assessment, documentation support, and a community mobiliser who is trusted enough to be let into a kitchen. Almost all of it recurs monthly. Almost none of it produces a plaque.
This is also why we encourage supporters to read an organisation's accounts rather than its photographs. We have written separately about how to read an NGO's financials and see where the money actually goes, and our own audited statements are published on the financials page of this site.
Measuring progress without flattering ourselves
Three numbers get used interchangeably in this sector and they are not the same thing.
- Enrolment is a name on a register. It is the easiest number to grow and the easiest to misread. A child can be enrolled and not have attended since August.
- Attendance is presence, and it is the number that tells you whether your operational fixes are working. We track it per child per day, not as a monthly average, because averages hide the girl who has quietly stopped coming.
- Learning outcome is whether she can read and calculate better than she could six months ago. It is the only number that matters to her life, and the only one that can go down.
We measure all three and we report the gaps between them, because the gap is the diagnosis. High enrolment with poor attendance is a household-logistics problem. Good attendance with flat learning outcomes is a teaching problem and it is ours to fix, not the child's.
Two other measures we have found worth the trouble: transition rate from Class 8 to Class 9, tracked child by child through the following academic year, and retention at twelve and twenty-four months. Anyone can produce a good first-quarter number. The second year is where programmes are actually tested.
Current cohort figures for the programme are here: [Setu to insert: number of girls enrolled in Project Uday in 2025-26], with a Class 8 to Class 9 transition rate of [Setu to insert: transition rate percentage and reference period].
What we would tell another organisation starting this
- Budget for recurring costs first and capital last. If your funding only covers assets, you will run a well-equipped centre with falling attendance.
- Hire from the settlement. Trust is the scarce input and it cannot be recruited from outside.
- Assess learning levels at intake, honestly, and teach to the level you find.
- Track individual children, not cohort averages. The average is fine right up until a third of the group has left.
- Treat documentation support as core programme work, not administration.
- Plan the exit. A girl who finishes Class 12 with no pathway into work or further training is a partial result, which is why our education work connects into the SWEEP employability programme and into the wider skilling work we describe in our account of skilling people with special needs.
Where you come in
The most useful support for work like this is unrestricted and multi-year, because the costs that keep a girl in school are monthly costs and the results take three to five years to show. If you are with a company weighing a CSR partnership, our guide on how to choose an NGO partner under Section 135 sets out what to ask for before you commit.
If you have subject expertise in mathematics, English or science and can commit to a fixed weekly slot for at least six months, remedial teaching is where volunteer time converts into measurable learning gain. Short, irregular volunteering does not, and we would rather say so. Details are on the volunteer page of this site.
To fund a specific component of the programme, or to discuss a multi-year commitment, get in touch with the team, or support the work directly through our donations page. We will tell you what the money buys, and we will tell you what it does not.
