Intersectional sampling in health research: A scoping review of approaches and structural constraints

Intersectionality, rooted in Black feminist praxis, holds that systems of power configure unique positions of disadvantage and privilege irreducible to single-axis analysis. Quantitative methods to estimate intersectional effects have advanced rapidly, but less attention has gone to a prior question: how samples are designed to make intersectional positions analytically visible. Prior systematic reviews have catalogued analytic methods without examining sampling as a distinct methodological object.

We conducted a scoping review following JBI guidance and PRISMA-ScR reporting standards, searching seven international and three regional databases (AJOL, LILACS, Global Index Medicus) from inception to February 2026, restricted to English-language publications. Screening used machine-assisted prioritization (ASReview); two reviewers conducted full-text screening, charting, and appraisal.

Of 51,154 records, ten studies (<0.02%) met inclusion criteria. All ten were conducted in the United States, United Kingdom, or Canada; none in low- or middle-income countries. This absence is consistent with prior reviews of the indexed, English-language methodological literature. We propose two provisional analytic instruments: a diagnostic framework (the Triad) and a four-category Taxonomy. Cross-cutting patterns included an implementation gap, a probability-intersection divide, a dimensional ceiling, and a recurring audit pattern.

Intersectional sampling in the indexed methodological literature remains rhetorical rather than operational. The Triad and Taxonomy are bounded starting tools, not global standards; their transnational application requires LMIC-led methodological reconstitution, not validation. We hypothesize five candidate mechanisms for the LMIC absence: infrastructure dependency, missing identity variables, digital access requirements, methodological hegemony in global health, and cell sparsity at population scale. These are to be tested empirically rather than treated as findings. Sampling designs that operationalize only marginalized positions are methodologically incomplete and should be reconsidered against intersectionality's bidirectional analytic of power.