The study context was secondary school enrolment in rural Rajasthan, specifically the transition from upper primary to secondary — classes 8 to 9 — where girls' dropout rates remain the most pronounced. Rajasthan's UDISE data consistently shows strong enrolment at upper primary level and a sharp fall at secondary, a pattern that has persisted despite significant investment in infrastructure, conditional cash transfers under the Mukhyamantri Rajshree Yojana, and state-level campaigns on girls' education.
The study asked parents why.
Survey data, collected first, produced a clear and largely positive picture. The overwhelming majority of surveyed parents in all three study blocks said they supported their daughters continuing to secondary school. Most said they intended for their daughters to complete class 12 and preferably pursue higher education. Very few cited financial cost as a barrier. The support for girls' education was, in stated terms, close to universal.
The enrolment data showed something different: actual dropout rates at the class 8-9 transition in the same blocks were between 28% and 35% depending on the block, consistent with the state pattern. The gap between what parents said they intended and what was observable in enrolment records was not marginal.
Focus groups were the instrument that made the gap legible. In group settings, the conversation moved differently than in household surveys. Individual preference statements — "I want my daughter to study" — were supplemented by accounts of how those preferences played out in practice, and particularly by accounts of what happened when individual preferences encountered community-level expectations.
The pattern that emerged was not parental deception in the survey. Most parents probably did hold the preferences they expressed. The issue was that individual preferences were being overridden by collective norm enforcement at the community level. Secondary schooling in these blocks typically required daughters to travel to a school that was not walkable from home — in most cases, 5 to 12 kilometres. This journey required either a bus or a vehicle, involved crossing areas without continuous social supervision, and placed girls in mixed-gender public spaces for the duration. In communities where a girl's safety and a family's reputation are understood to be tightly linked, the decision to send a daughter on that daily journey is not made by the household alone. It is evaluated and, in some cases, sanctioned by the extended family and immediate community.
Families who sent daughters to secondary school despite this social environment described an informal cost: monitoring, commentary, and in some cases active pressure to stop. Several families described withdrawing their daughters not because of a specific incident but because sustained social friction made the decision unsustainable. Others had anticipated this cost and never enrolled in the first place.
This is the mechanism that individual-level surveys cannot easily detect. A household survey that asks about preferences and barriers treats the household as the unit of decision-making. In communities with strong norm-enforcement mechanisms, the household is not the relevant unit. The decision is made in a social field that extends beyond the household boundary, and neither the preferences nor the constraints exist at the individual level alone.
For programme design, the implication is specific: infrastructure investments — schools, buses, hostels — are necessary but not sufficient. The binding constraint in these communities is not that parents don't want their daughters to attend secondary school. It is that they cannot sustain that decision against community pressure without social cover. Interventions that have worked in analogous contexts have provided that cover, either by making the norm-deviant decision more common (reducing individual exposure) or by creating community-level forums in which parents who want their daughters to continue could express and reinforce that preference collectively.
The Rajasthan finding is not novel. Duflo (2012) and subsequent work on social norms and female education document similar patterns across geographies. What it illustrates specifically for research methodology is that preference elicitation in community-embedded contexts requires instruments sensitive to the collective dimension of individual decisions.
UDISE+ Rajasthan district data on enrolment, dropout, and school location. ASER 2022 for state-level learning outcome context. Mukhyamantri Rajshree Yojana administrative data from the Rajasthan Directorate of Women Empowerment.