Studying Domestic Violence in a Mumbai Chawl
A doctoral student plans a survey of domestic violence among 400 households in a Mumbai chawl. Residents are 60% Marathi-speaking, 25% Hindi-speaking, and 15% Urdu-speaking. A local mahila mandal agrees to help but wants to see the questionnaire and be present during interviews. The police station refuses to share crime data. A men's group has circulated a message that the survey is 'anti-family'. The student has eight months of funding, one male research assistant, and no prior contact with the chawl. The university ethics committee asks for a protocol before fieldwork. The chawl has one narrow lane and shared toilets, so privacy is a practical problem. The student is under pressure to produce statistically generalizable findings but also wants to understand why women do or do not report violence.
1. Choose and justify a research design, including sampling and language strategy. Why is a pure survey inadequate here?
Read the worked answer
A mixed design is best, but the priority is safety, not generalisability. A pure survey of 400 households would produce numbers but may miss the reasons women do not report, and it would expose interviewers and respondents. Use a two-stage design: (1) a short household survey with a random sample of 200 households, stratified by language, with anonymous response cards; (2) 25 in-depth interviews with women who volunteer, drawn through the mahila mandal, plus 8 interviews with men. The sample frame can be the chawl's electricity list or ration cards, not police data. Translate and back-translate the questionnaire into Marathi, Hindi, and Urdu. Use female interviewers for the violence module; the male assistant can do mapping and logistics. The claim to generalisability must be limited to the chawl. The reasoning: domestic violence is a sensitive topic where disclosure depends on trust and privacy; a survey alone gives prevalence, ethnography gives meaning, and mixed methods let you check whether non-reporting is a measurement artefact.
2. Set out the ethical protocol, especially for disclosure of violence and mandatory reporting.
Read the worked answer
Ethical protocol: (1) informed consent in the respondent's language, with a clear statement that participation is voluntary and that she can skip questions; (2) privacy-interview in a room with no husband or mother-in-law present, and use a code, not a name; (3) a safety protocol: if a woman discloses current danger, the interviewer does not promise secrecy she cannot keep. The team should carry a list of local services (One Stop Centre, 181 helpline, shelter) and ask the woman what she wants. Mandatory reporting is not automatic for adult domestic violence in India; the researcher should not file a police complaint without consent unless there is an immediate threat to life or a child is at risk. (4) The mahila mandal's presence must not become surveillance; it can introduce the team but should not sit in on interviews. (5) Debrief and referral for interviewers, who may hear traumatic accounts. The reasoning: confidentiality is qualified by safety, but over-reporting can expose women to retaliation.
3. How would you handle the mahila mandal as gatekeeper and the men's group's resistance?
Read the worked answer
Treat the mahila mandal as a gatekeeper with its own interests, not as a neutral entry point. Negotiate a written agreement: they see the questionnaire, but they do not see completed responses and do not sit in interviews. Offer a community feedback session at the end, not raw data. For the men's group, avoid a public debate; instead, meet two or three influential men and explain that the study covers men's health and household stress too, and that no individual will be identified. Use the eight months to do mapping and 10 pilot interviews first. If resistance continues, shift to a smaller ethnographic sample and be transparent about limits. The judgement is that access is political; a researcher who forces entry may get unreliable data and harm respondents. Also, police data refusal is not a fatal flaw-it is itself a finding about reporting.