Intersectionality (Fairness)

Intersectionality in the context of fairness refers to the analytical lens that recognises how multiple protected attributes—such as race, gender, disability, sexual orientation, or socioeconomic status—can combine to produce unique patterns of discrimination. Rather than treating each attribute independently, intersectional thinking asks how the overlapping identities of a person shape their experience with bias, often amplifying inequities in ways that are not predictable from any single factor alone.

The concept matters because policies, algorithms, and decision‑making systems that only address one dimension of fairness can miss or even worsen hidden harms. For example, a hiring tool calibrated to remove gender bias might still disadvantage women of color if the underlying data reflect compounded disparities. By explicitly accounting for intersecting identities, designers can craft interventions—such as adjusted thresholds, richer feature representations, or targeted data collection—that better protect those at the margins.

Intersectionality shows up wherever fairness is operationalised: in legal frameworks that assess disparate impact across multiple groups, in social science research measuring health outcomes for overlapping demographic categories, and increasingly in machine‑learning pipelines that aim to audit, mitigate, or explain bias. Practitioners use it to guide dataset audits, to define more nuanced protected classes, and to evaluate whether a system’s performance gaps persist after single‑attribute corrections.