Why the Dropout Rate Doubles When You Leave the Office
A sociology student in a Bihar district is asked to study why girls leave secondary school. The district education office reports a dropout rate of 6% for girls in Classes IX–XII. A local NGO working in the same blocks claims the real figure is 23%. The district has 40 villages and about 12,000 households. School registers are incomplete: many entries read 'not known', and some girls are enrolled but have not attended for months. The student has three weeks, one research assistant and ₹25,000. A sarpanch offers to introduce 'cooperative' respondents and says he will 'take care of' any difficult families. On the first visit, only 60% of enrolled girls are present; some are married and living elsewhere, and several boys in the same classes have also left. A teacher who is also a respondent teaches in one of the sample schools.
1. Define 'dropout' operationally and choose a research design within the constraints. Justify your sampling.
Read the worked answer
First, operationalise: dropout is a girl aged 12–18 who was enrolled in Classes VI–XII in the district at the start of the academic year but has been absent for 30 consecutive school days without a transfer certificate, or whose name is on the roll but who has not attended for 45 days. This separates never enrolled, dropout, migration and temporary absence.
Within three weeks and ₹25,000, a census is impossible. Use a two-stage cluster sample: randomly select 8 of the 40 villages after stratifying by distance from the block town and by caste composition; in each village, list all households with girls aged 12–18 (about 300), then randomly select 40 households - 320 girls in total. Add an attendance-register audit in the 8 villages and 15 short interviews with girls who left.
The sarpanch's 'cooperative' list must not be the sampling frame; it will over-represent families with good relations with the panchayat. Random selection and a household listing give every girl a known chance of inclusion. Triangulate school records, household reports and interviews. The design is feasible: two fieldworkers, eight villages, four days each, with time for data cleaning.
2. How would you handle the gap between the official 6% and the NGO's 23%, and the sarpanch's offer to introduce 'cooperative' respondents?
Read the worked answer
Treat the gap as a measurement question first, not as a dispute about honesty. Three sources of difference are likely. First, definition: the official figure may count only girls who formally withdraw, while the NGO may count girls who are enrolled but absent, married, migrated or working. Second, denominator: the official rate may use enrolled girls, the NGO may use all girls in the age group. Third, recording: 'not known' entries and incomplete registers hide dropout. The study should report all three rates separately - official, adjusted and self-reported - and explain the difference.
On the sarpanch: refusing the offer may close doors, but accepting his list would introduce selection bias. Use him for access and logistics, not for selecting respondents. Ask him to announce the study, then do a random walk or use the school register to list households; interview without him present. If he insists on accompanying the team, record it as a limitation.
The NGO figure should be treated as a hypothesis to test, not as the true rate. A good answer shows that triangulation is not averaging the two numbers; it is explaining why they differ.
3. Identify three ethical risks and explain how you would manage them.
Read the worked answer
Consent and power: girls and parents may agree because a teacher or sarpanch is present. Manage by explaining the study in Magahi or Hindi, making clear that refusal has no effect on school or welfare benefits, and interviewing in a private space. For minors, take parental consent and the girl's assent.
Confidentiality: a girl may disclose early marriage, work or harassment. Do not record names in the dataset; use codes; keep the consent sheet separate; do not report village-level findings that could identify a family.
Safety and expectations: respondents may expect money or help after the survey. Do not promise benefits; keep a referral list for child protection, health and legal aid, and explain that the study cannot guarantee assistance. Also avoid raising the dropout issue in front of teachers who may penalise the girl.
A fourth risk is researcher safety in remote villages; plan travel in daylight and share the itinerary. Ethical review is not a formality: it shapes the sampling, the wording of questions and the decision not to publish identifiable vignettes.