The Sarpanch's Guest: Access and Bias in a Village Voting Study
A postgraduate student from Jaipur is studying how caste influences voting in a village of 2,400 voters in Nagaur district. She has three weeks, one research assistant, and no prior contact. The sarpanch offers to introduce her to 'important people' and to arrange a meeting of the village council. Her first 30 interviews, conducted in the presence of the sarpanch's nephew, show almost no mention of caste; respondents say they vote on 'development issues'. Her questionnaire asks respondents to tick their caste from a list, but many refuse. Women, who are 48% of voters, are rarely available during the day and are answered for by their husbands. She must submit a fieldwork report that her supervisor will assess for methodological soundness.
1. Identify two threats to the validity of her findings and explain the likely direction of each bias.
Read the worked answer
Two threats stand out.
1. Gatekeeper-induced response bias. The sarpanch's nephew is present during interviews. Respondents know the sarpanch can influence local benefits, so they give socially safe answers. The likely direction is under-reporting of caste voting and over-reporting of 'development issues'. This is not random error; it is systematic, and it biases the main variable the study wants to measure.
2. Sampling and proxy-respondent bias. Women are 48% of voters but are reached mainly through husbands. The sample therefore over-represents men and households favourable to the sarpanch, and women's preferences are recorded second-hand. The likely direction is to make the village look more homogeneous and more male-dominated in political reasoning than it is.
A third, related problem is the caste question itself. Refusal to tick a caste box does not mean caste is irrelevant; it may mean the question is perceived as intrusive or dangerous. Treating refusal as 'no caste effect' would confuse non-response with substance.
Implication: the 30 interviews cannot be treated as a representative baseline. They are a convenience sample shaped by patronage and gender. The student should report them as exploratory, not as evidence that caste does not matter.
2. Redesign her access strategy and instrument so that the three-week fieldwork can still produce credible evidence. Give concrete steps.
Read the worked answer
A workable redesign in three weeks:
- Break the single gatekeeper. Request permission from the sarpanch for the study, but do not let his relatives accompany interviews. Recruit a local assistant who is not a member of the sarpanch's household, and rotate assistants across hamlets.
- Map the village first. Use voter lists and hamlet maps to list households by locality and caste cluster, without asking individuals to tick a caste box. Caste can be inferred at the hamlet level and later validated in interviews, which lowers refusal.
- Change the instrument. Replace the direct 'tick your caste' item with open questions about how candidates are chosen, who visits households, and what people discuss at meetings. Ask about the last three elections concretely rather than about 'caste influence' in general.
- Reach women directly. Schedule interviews in late morning or early afternoon, use a woman assistant, and hold small group discussions in courtyards rather than mixed meetings.
- Build in triangulation. Compare interview accounts with polling-station-level turnout, candidate profiles, and observations at two or three campaign events. If accounts and observed behaviour diverge, report the divergence rather than smoothing it over.
- Keep a field diary. Record who was present, who refused, and how access was negotiated. This turns access problems into analysable data.
Credibility in three weeks comes from transparency about limits, not from pretending the sample is representative.
3. Should she accept the sarpanch's help? Discuss the ethical and methodological trade-offs.
Read the worked answer
Accepting help is not automatically unethical, but uncritical acceptance is. The trade-off has three parts.
Access versus independence. The sarpanch can open doors that would otherwise take a week to open. Refusing all help may mean no access at all in a short fieldwork window. But accepting his nephew as a constant presence converts the study into a patronage visit.
Informed consent. Respondents must understand that participation is voluntary and that refusal will not affect their access to village services. If the sarpanch is present, that assurance is hard to believe. Consent obtained in front of a local power-holder is compromised consent.
Reciprocity and risk. The student should not promise benefits she cannot deliver, and should not publish identifiable accounts that could expose respondents to retaliation. Anonymising hamlets and using composite quotations is a minimum safeguard.
A defensible position: accept limited, public help to enter the village, disclose the purpose of the study to the council, and then conduct interviews independently. Record the help in the field notes. The ethical test is not whether a powerful person helped, but whether respondents could refuse and whether the report lets readers see how access shaped the evidence.