Invited Commentary: Comparing Approaches to Measuring Structural Racism

Abstract

In their article, Adkins-Jackson et al. (Am J Epidemiol. 2022;191(4):539–547) provide much needed insight on current approaches and challenges to epidemiologic research on structural racism. The authors encourage researchers to consider how structural racism is conceptualized in extant and future work, and whether existing measures accurately reflect the multidimensional nature of structural racism. In the spirit of continuing this conversation, I expand upon approaches to measuring structural racism. I compare the merits and challenges of scales, indexes, indicators, and policy-based measures of structural racism for use in epidemiologic research.

Editor’s note:  The opinions expressed in this article are those of the authors and do not necessarily reflect the views of the American Journal of Epidemiology.

Recent calls from professional organizations (1, 2), the National Institutes of Health (3), and academic journals (4) demonstrate a growing demand for epidemiologic research on structural racism as a determinant of racial health inequities. These efforts aim to redress the historical marginalization of research on structural racism in public health pedagogy, funding institutions, and academic journals (5). Increasing interest in the epidemiologic study of structural racism requires that researchers grapple with existing and new measures of structural racism. In this issue of the Journal, Adkins-Jackson et al. (6) provide foundational guidelines for study design and measurement. They challenge researchers to consider how structural racism is conceptualized, measured, and modeled, in which populations and at which point in the life course—questions that promote nuanced and meaningful work.

In this commentary, I add to the conversation started by Adkins-Jackson et al. by comparing different measurement approaches to operationalizing structural racism in epidemiologic research. The authors recommend the use of structural equation modeling and factor analysis to develop scales of structural racism that reflect the joint influence of structural racism across multiple levels and institutions. This is indeed a useful approach, and it provides a clear methodological answer to the call from Gee and Ford (7), nearly a decade ago, for public health researchers to study the multidimensional aspects of structural racism. I contrast this approach with other measures not to offer one method as preferable to another, but rather to aid epidemiologists in study design decisions by laying out a range of measurement tools that have different interpretations and utility across various research contexts.

SCALES, INDEXES, AND OTHER MEASURES OF STRUCTURAL RACISM

There is no scientific consensus on how to best measure structural racism, and no single measure can fully represent the totality of ways that structural racism operates. Therefore, it is incumbent upon researchers to compare findings across different measures, triangulate evidence, and form hypotheses about how structural racism affects health outcomes. Metaphors can be useful in these endeavors (8). Gee and Hicken (9) used the analogy of a buckminsterfullerene to describe the multidimensional nature of structural racism as a soccer ball-like molecule made up of connections between carbon atoms that represent different societal institutions. It is these multiple, interlinked institutions that differentiate structural from institutional racism (a term that describes racism as it arises in particular institutions). The latent construct approach preferred by Adkins-Jackson et al.—using scales to extract an underlying measure of structural racism—is a way for researchers to observe larger sections of the buckminsterfullerene, empirically describing the connections between several domains (groups of connected metaphorical carbon atoms). Arguably, equally important to describing which domains vary together is producing work that examines interinstitutional linkages, or the bonds between carbon atoms and the conditions that allow them to hold. Below and in Table 1, I describe additional measurement approaches that are useful in these efforts.

Table 1.

Methods for Measuring Structural Racism

Measure	Example Articles	Strengths	Limitations
Scales	(25–27)	Reflects the multidimensional nature of structural racism and joint effects of several dimensions	Less direct implications for intervention
Indexes	(13)	Reflects the multidimensional nature of structural racism and joint effects of several dimensions	Sum scores assume equal weight of each indicator; only a few indicators can be modeled simultaneously
Indicators	(10–12, 28, 29)	Most easily interpreted and implemented	Describe the consequences of structural racism but do not directly measure its multidimensional nature
Specific policies and groups of policies	(30–32)	Direct implications for intervention; can measure the mechanisms of structural racism’s effects on health	Insufficient variation in exposure for deeply entrenched policies and policy areas

Indicators and indexes

Several studies have measured structural racism using single or multiple contextual indicators and/or racial inequities in these contextual indicators, including racial inequities in rates of imprisonment, unemployment, and educational attainment, as well as measures of racial residential segregation (10–12). Assessing the relationship between a single contextual indicator of structural racism and a health outcome is relatively straightforward and, in comparison to scales, uses units that are directly interpretable. Multiple indicators can be summed or otherwise aggregated to construct indexes of structural racism. For example, Homan et al. (13) developed a measure of “structural intersectionality” to conceptualize and measure intersecting systems of oppression (e.g., racism, sexism, ableism, etc.) using multiple contextual indicators that were then integrated in a combined additive measure. However, indexes that sum multiple indicators assume that each indicator should be equally weighted, and it can thus be useful to examine relationships for indicators separately (10). Examining interactions between two theoretically relevant indicators may glean useful insights for understanding how structural racism operates, although positivity violations limit how many contextual indicators can be modeled simultaneously (14).

Indicators are the result of processes that involve several institutions and dimensions of structural racism, even if these mechanisms are not explicitly modeled. For example, racial inequities in rates of imprisonment reflect racially inequitable laws, police surveillance, judicial practices, and other factors. Researchers should adequately contextualize their findings from indexes and indicators to describe which institutions and domains of structural racism are being detected by the study’s indicators. Theory should be used in interpreting indicators because it helps researchers consider what unmeasured factors associated with the indicators could be contributing to the results (15–17). Perspectives from outside of the academy can also guide the development of new indicators and indexes (as well as scales). For example, Chambers et al. (18) partnered with women’s health organizations to conduct focus groups with Black women, the results of which both validated existing theories and identified unique domains of structural racism.

Policies and groups of policies

Policies can reinforce racism across multiple institutions (19). Despite their importance, the epidemiologic study of state-sanctioned racism in the form of laws and policies is an emergent but still small area of research (15). To address this gap, Agénor et al. (20) recently developed a database of structural racism-related laws for health equity researchers. These laws reflect structural racism because they “determine the inequitable allocation of social, economic, political, and environmental resources and harms across racial/ethnic groups” (20, p. 429). Tools from legal epidemiology, such as legal mapping and econometric methods, can also be useful for measuring exposure to structural racism in policies (21). These methods can additionally be applied to evaluate policies aimed at addressing structural racism, and assessing whether these policy changes narrow racial health inequities.

As with indicators, policies reflect processes that bridge institutions and can affect health outcomes through multiple pathways. For example, immigration policies are hypothesized to generate health inequities because they exclude specific groups from the range of rights and privileges that citizenship affords, including political and civic representation, legal protections, and public benefits (22). If the ultimate purpose of epidemiologic work is to intervene upon population distributions of disease (8), then measuring structural racism using policies is particularly informative. This approach can also be useful for identifying racial health inequities that arise from race-neutral policies that nevertheless perpetuate racialized harms (23).

However, some methods can be used only to assess health consequences of observed, politically feasible policy changes, as opposed to more transformative policies that have not been enacted (24). Because racially inequitable policies and groups of policies are often deeply entrenched, they lack the temporal variation in exposure needed for comparisons. Indeed, even incremental reforms to racially inequitable policies may not be sufficient to observe expect ed improvements in racial inequities in health outcomes at the population level. As Adkins-Jackson et al. note, a policy improvement (e.g., ending redlining) can be followed by a similar policy or practice that persists over time (e.g., racial bias in home lending), thus limiting the observed variation in that type racially inequitable policy (6).

CONCLUSION

The task at hand for epidemiologists studying structural racism includes not only developing and refining measures but also evaluating their validity, and ultimately, understanding how they (re)produce racial inequities in health outcomes in order to advance equitable change. Adkins-Jackson et al. (6) are charting new paths forward in this work by encouraging epidemiologists to consider how measures of structural racism can incorporate multiple institutions and domains, among several other useful recommendations. The measurement approaches described in this commentary generate different kinds of insights into the relationship between structural racism and health, and I compare these approaches to invite further epidemiologic inquiry and conversation as researchers continue to investigate this fundamental driver of racial health inequities.