Six Weeks in the Resettlement Colony
An NGO in Bhubaneswar commissions a rapid assessment of a resettlement colony of 380 households displaced from a reservoir project in the 1990s. The budget is ₹4.2 lakh and the deadline is six weeks, before a funder's review. The team is a principal investigator, two postgraduate students, and a local key informant who is also an office-bearer of a political party. Many residents are non-literate in Odia, several lack title documents, and some are recent arrivals from another district who fear that any enumeration will bring them to official attention. The ward councillor offers to 'select 40 representative families' and to sit in on the interviews. The funder asks for photographs of daily life and for the names of participants in an annexure, and separately pressures the NGO to release provisional findings two weeks before the deadline to support a parallel grant application. The team must decide the ethics, the sampling and the reporting plan before fieldwork begins.
1. Set out the consent, confidentiality and conflict-of-interest decisions the team must take before the first interview, giving your reasoning for each.
Read the worked answer
Consent with non-literate participants. Written forms are not the standard here; oral consent, recorded with permission or witnessed by someone not connected to the NGO, is. The explanation must be in Odia and in the residents' own language, must state that refusal has no effect on any entitlement, and must say what will happen to the findings. Consent must be sought from each respondent separately, not from a household head on behalf of a spouse or adult children.
The councillor. Selecting respondents and sitting in on interviews would make the sample an artefact of patronage and would suppress criticism of local services. The team should decline both, in writing, while keeping the councillor informed through a briefing at the end. Her role in endorsing the study can be acknowledged without letting her presence shape answers.
The key informant. A party office-bearer is a gatekeeper with an interest in which grievances are heard. Retain her as a guide to the colony's layout, but: (a) do not let her recruit respondents, (b) brief her in writing on not attributing statements, (c) have a second, unaligned guide for the other side of the colony, and (d) never conduct interviews through her translation alone.
Confidentiality and photographs. Names in an annexure plus photographs of identifiable houses can combine to expose undocumented residents to precisely the risk they fear. Recommend: no names, household codes only; photographs of the colony with no faces and no house numbers unless written consent; storage of the recording device's files on one encrypted drive; and a rule that no individual interview is ever replayed to the NGO staff who work in the colony.
Approvals. Even a student rapid assessment should go to an institutional ethics committee, and the team should agree in writing, before fieldwork, on what it will not do if asked.
2. Given six weeks and ₹4.2 lakh, how would you build a defensible sample? State the design and the likely biases you would document rather than try to hide.
Read the worked answer
Design.
- Household listing first (2 weeks). A complete spatial listing of all 380 structures, with a sketch map and GPS points, is the single most valuable product and is achievable in six weeks. Number every structure; record vacant, non-residential and shared units.
- Stratify by obvious heterogeneity. The colony is not one population: long-settled titleholders, landless tenants, recent arrivals from the other district, and female-headed households are the strata that matter for the study questions. Allocate the sample across strata rather than proportionally, with a minimum of 25 per stratum.
- Random selection within strata using the listing, replacing absent households by pre-drawn substitutes, not by the fieldworker's convenience. Target 150–180 households for the survey; this is enough for stratum-level percentages with wide but stated confidence.
- Purposive sub-sample for qualitative work: 20 life histories chosen to include elderly residents, recent arrivals, widows and households resisting enumeration, plus 2 group discussions with women, held separately from men.
- Deliberate recruitment of hidden cases via snowball from multiple starting points, never only the councillor's list.
Documented biases. Non-response will be highest among recent arrivals, which is exactly the group whose vulnerability matters most; the report must state the refusal rate by stratum. Recall of 1990s displacement will be shaped by later experience. The absence of a sampling frame for households sharing a single structure will undercount the poorest tenants.
What the budget cannot buy. A full census, a socio-economic baseline by income, or follow-up. The report should say so and recommend a second round rather than imply that a six-week rapid appraisal has produced a baseline.
3. How should the team handle the funder's two requests - photographs with names, and early release of provisional findings? Draft the operative decisions and the wording you would put in the report.
Read the worked answer
Photographs and names. These are not clerical requests; in this colony they are a confidentiality risk. Decision: photographs only with separate, specific consent for each image and each use; no identifiable images of individuals in any report; no names in any annexure, appendix or spreadsheet shared outside the core team. Wording for the report: 'Identifying details of participants are withheld because several households fear that public association with this study could affect their dealings with local authorities. Household codes are used throughout, and the code key is held only by the principal investigator.' If the funder objects, the team should offer a substituted product - photographs of infrastructure, lanes and water points, and composite profiles - rather than trade away consent.
Early release of partial findings. Provisional results released for a grant application, without the caveats and without the community having seen them, will circulate as settled fact. Decision: (a) no release before the data are checked and the strata weighted; (b) if the funder needs something in writing, supply a short 'method and emerging themes' note that quotes no respondent and states explicitly that findings are unvalidated; (c) the same note goes to the colony's residents' committee first.
Clause to insert. 'Preliminary material may not be cited, reproduced or used in support of any funding application without the written agreement of the principal investigator and a statement of its provisional status.'
Telling the NGO no, constructively. Frame the refusal around risk to the NGO, not only to respondents: a name list that leaks harms the organisation's ability to work in the colony for years. Offer the alternative product in the same conversation, so that the answer is 'not this, but here is something usable'.