The Shift From Individual to Structural

In the decades following the Civil Rights Act of 1964 and the Voting Rights Act of 1965, the most visible forms of explicit racial discrimination in the United States — segregated schools, whites-only hiring notices, literacy tests at the ballot box — became illegal and, over time, substantially less common. Survey measures of openly expressed anti-Black prejudice among white Americans declined steeply across the second half of the twentieth century (Schuman, Steeh, Bobo, and Krysan 1997, pp. 104-110). On a purely attitudinal account, one might have expected racial disparities in wealth, health, education, and criminal justice to narrow correspondingly. They did not. On several key indicators, disparities stabilized; on others, they widened.

This empirical pattern — declining individual-level prejudice alongside persistent or growing group-level inequality — forced a conceptual reworking. If racism is understood only as the conscious animus of individuals, then the persistence of disparity is puzzling and easy to explain away as the residue of personal choices or cultural differences. If racism is understood structurally, the persistence is not puzzling at all: the institutional arrangements that produced the disparities were not dismantled when the explicit legal framework was, and those arrangements continue to operate even when staffed by individuals who hold no conscious racial animus.

The vocabulary of institutional racism was coined by Stokely Carmichael (later Kwame Ture) and Charles Hamilton in Black Power (1967). They distinguished individual racism — overt acts by identifiable actors — from institutional racism, defined as the routine operation of established and respected institutions in ways that produce racially disparate outcomes (Ture and Hamilton 1967, pp. 4-5). The distinction was consequential: institutional racism does not require individual racists to operate. It requires only that the institutional procedures — mortgage underwriting criteria, school district boundaries, sentencing guidelines, hiring networks — produce systematically different outcomes for racially defined groups.

Eduardo Bonilla-Silva (1997) formalized a related argument in American Sociological Review. The proper object of analysis, he argued, is the racial structure — the totality of social relations and practices that reinforce racial privilege. In this framing, racism is not primarily an ideological error but a set of material arrangements: the racial structure generates racial ideologies and racially patterned behavior, not the other way around (Bonilla-Silva 1997, pp. 469-473). This shift parallels a more general move in sociology from taking beliefs as causes to taking them as effects of structural position.

A third strand came from medical sociology and social epidemiology. Jo Phelan and Bruce Link (2015) proposed that race, like socioeconomic status, functions as a fundamental cause of health inequalities: an enduring social characteristic that affects access to flexible resources (money, knowledge, power, prestige, beneficial social connections) and therefore shapes exposure to a wide range of intervening risk factors. Interventions that target a single intervening mechanism — a specific disease, a specific behavior — may narrow that particular gap while the underlying fundamental cause generates new disparities on other outcomes (Phelan and Link 2015, pp. 311-314). Race-as-fundamental-cause is a structural claim: the mechanism is the allocation of flexible resources across a racialized population, not the particular pathology under study.

Structural racism, as the term is used in contemporary sociology, integrates these strands. It names the cumulative, compounding, and often impersonal institutional arrangements — across housing, education, labor markets, credit, health care, and criminal justice — that reproduce racial disparities even in the absence of individual racist intent. The framework does not deny that individual prejudice exists or matters; it insists that the individual-level variable is insufficient to explain the population-level pattern. In the four domains examined in this topic, the insufficiency is empirically clear: disparities are large, persistent, and patterned in ways that no plausible distribution of individual attitudes could generate on its own.