Designing an Intersectional Study of Street Vendors
An interdisciplinary team is designing a study on eviction and women street vendors in a Tier-II city. The initial plan is a survey of 300 women vendors, with analysis limited to gender. The funder asks why caste and religion are absent; one team member replies that intersectionality is inherently qualitative and cannot be quantified, while another proposes adding caste and religion to a regression with three-way interaction terms. Municipal ward data show that 62% of women vendors are OBC, 21% SC and 9% Muslim, and that eviction notices in the last year were concentrated in two of eleven market zones. The budget supports 40 qualitative interviews; the ethics committee meets in three weeks; no baseline list of vendors exists. The vendors' association is male-dominated and has not shared its membership list. The team must decide what counts as a defensible intersectional design rather than a gesture.
1. The team asks you to adjudicate between the qualitative-only and add-interaction-terms positions. What is your recommendation, and what makes it intersectional rather than merely disaggregated?
Read the worked answer
Neither position is adequate. The qualitative-only claim treats intersectionality as a method rather than an analytic sensibility, while the interaction-term proposal assumes that adding caste and religion to a regression automatically captures intersecting power. With 300 cases, three-way interactions would produce sparse cells and unstable estimates, and the missing sampling frame already biases who appears.
A defensible design has four elements. First, build a sampling frame through a market census in all eleven zones, stratified by zone and by self-identified caste and religion categories. Second, report outcome patterns across intersectional strata descriptively, not as one coefficient. Third, use the 40 interviews purposively to trace mechanisms: how municipal officials, caste networks, male traders and family responsibilities shape who receives notices and who can resist. Fourth, analyse eviction logs and municipal documents to link micro experiences to meso-level governance.
The design is intersectional because it asks how gender, caste and religion jointly structure exposure and response, not because it multiplies variables. State the limits clearly: small cells, self-report, and no claim of national representativeness. The test is whether the design can show differential processes, not whether it contains an interaction term.
2. The funder requests a single intersectionality index to rank the eleven market zones. Explain the problems with this request and propose an alternative reporting strategy that remains useful to the municipal authority.
Read the worked answer
A single index would force different axes of inequality into one scale, hide the mechanisms that produce eviction, and create false precision. It would also encourage a ranking of zones that could stigmatise vendors and misdirect resources. Intersectionality is not a score; it is an analysis of how power relations combine.
An alternative is a disaggregated evidence matrix. Rows can be intersectional strata (for example, SC women in zone A, Muslim women in zone B, OBC women with young children), and columns can be separate indicators: exposure to eviction notices, access to legal aid, security of tenure, income loss, and presence of a women's collective. Present confidence intervals where numbers are small and use qualitative typologies to explain the patterns. Add a mechanism section showing how ward officials, market associations and caste networks operate in the two high-eviction zones.
The municipal authority can use this matrix to target enforcement and support without pretending that one number captures intersectional disadvantage. If a composite is politically unavoidable, publish its components and weights, disaggregate it, and state what it cannot measure.