Incorporating a more expansive theory of racism into child and family policy systems

Abstract

Berger and Harden have offered a comprehensive and compelling overview of the state of empirical research on racial inequalities in child welfare system outcomes. They show that Black children and families experience CPS intervention across a range of outcomes more frequently than White peers. They suggest four causal pathways that could account for these inequalities: 1) differential surveillance; 2) decision-maker bias within CPS; 3) differential risk of child maltreatment; and 4) structural racism. They suggest that evidence for differential surveillance and within-CPS bias are likely small (or negligible) contributors to Black/White inequalities in child welfare outcomes; instead, current evidence strongly points to differential risk of maltreatment and structural racism as key drivers of differential Black child and family exposure to child welfare systems.

While Berger and Harden's set of four causal pathways do adequately capture those proposed in most research literature, this framework misses a subtle vector for how racism impacts the CPS policy process. As I attempt to illustrate in Figure 5 of my first essay, historical and contemporary racism have impacted the policy field itself. The quantity and quality of interventions available to front-line social workers to respond to children and families in crisis are themselves products of racist social processes. The narrow focus on marginal effects of race on CPS outcomes common in econometric analyses implicitly naturalizes these arrangements. From a critical perspective, the common counterfactual question may be more accurately stated as: “within a policy system known to have racist design features and policy goals, and in a context of deep structural inequality, do otherwise comparable Black and White children experience different outcomes?”

If we bracket our definition of the concept of racial discrimination to merely the marginal impact of the perception of skin color on a discrete outcome of a policy process we can obtain an answer to this question. However, these approaches reduce the complex social stratification system of race to the perception of phenotypical differences (Kohler-Hausmann, 2018), obscuring the structural and institutional causal pathways that produce race as a system of social and family stratification salient for CPS processes (Feely & Bosk, 2021). For example, in recent work, Baron and colleagues (2024) found evidence of differential treatment of Black children by CPS agency workers in Michigan but paradoxically argued that evidence points to an “under-protection” of White children compared to Black children through their assessment of the probability of intervention conditional on their measurement of maltreatment risk. While I disagree with their interpretation of this evidence, this finding does present an interesting and useful counterfactual that deserves further scrutiny.

What kind of child welfare system would we have if White children, rather than Black and Indigenous children, had been conceptualized by policymakers as the focal targets of intervention? The historical evidence on this question is compelling: Core legislation and institutions that form the foundation for the contemporary child welfare system were deeply influenced by racist political calculations, and systems engaged in punitive transformations when forced to accommodate Black children and families (Simmons, 2020; Spinak, 2023; Ward, 2012). For example, Simmons (2020) showed that the desegregation of New York City's child welfare infrastructure in the mid-20th century was enabled by and triggered a series of policy changes that funneled dependent Black children through public criminal justice and juvenile justice institutions, rather than the private welfare institutions that White children and families received services from. And Raz (2020) compellingly demonstrated how our current regime of mandated reporting, investigation, and family separation resulted from a series of anti-Black political maneuvers in the 1970s and 1980s. A deep body of scholarship has demonstrated that when Black children and families are framed as the targets of social policy interventions, policymakers tend to design and implement systems focused on social control, rather than amelioration (Quadagno, 1994; Soss et al., 2011). A form of anti-Black, anti-Indigenous, and often anti-immigrant welfare chauvinism (Andersen & Bjørklund, 1990) prevails in the United States that typically reserves supportive interventions for middle- and upper-class White families conceived as somehow virtuous, while directing investigative, punitive, and pathologizing interventions at the poor, who are framed as undeserving (Schneider & Ingram, 1993).

Our theories and methods for evaluating the impact of racism on policy outcomes must be expansive. Narrow econometric approaches bracket many of the most important forms of racism as exogenous to policy outcomes, and in so doing, drastically underestimate the impacts of racism. Policy analysts and scholars must take a more expansive view that appreciates how racism broadly structures the distributions of risks, the practices of bias, and policy systems themselves.

I am in full agreement with Berger and Harden that “there is no cheap and easy solution for reducing child maltreatment, CPS involvement, and racial disparities therein.” Our current set of policy systems targeted at addressing child maltreatment and family crisis are reactive, narrow, and incapable of solving centuries-old problems of racialized inequalities in family life and childhood. How then should agencies, policymakers, and policy scholars proceed in advancing an agenda that simultaneously advances child safety and equity?

As Berger and Harden note, racially unequal exposure to CPS triggers criticism precisely because of the coercive and disruptive modalities it employs (i.e., surveillance, investigation, separation) in efforts to advance child safety. Strictly ameliorative interventions do not generate the same kinds of ethical concerns. From a normative perspective, if we can achieve similar outcomes relative to child safety from supportive interventions when compared with coercive interventions, the supportive approach should be preferred. By shifting our focus, decomposing bias and risk as contributors to CPS outcomes becomes a secondary or tertiary concern for equity.

A subtle tweak to the common counterfactual for policy scholarship can help us move in this direction. Rather than comparing safety outcomes for children to a no-intervention counterfactual, we can evaluate routine CPS interventions (i.e., foster care) in comparison to an alternative (and abolitionist) counterfactual of adequate material support for families. From a design perspective, this is complex. But analytically, it helps us to address a central question: How would safety and well-being outcomes for children and families differ under a reactive and coercive system (i.e., CPS) as compared to a system broadly focused on material security and stability for families?

The evidence available on this revised question is promising (Ananat & Garfinkel, 2024; Pac et al., 2023). We know that income is a key cause of child safety and CPS intervention (Berger & Waldfogel, 2011; Cancian et al., 2013; Gibson et al., 2020; Moullin & Milne, 2024). And evidence on the impacts of COVID pandemic–era anti-poverty interventions like the child tax credit suggests that anti-poverty measures have dramatic and substantial positive impacts on safety (Campbell et al., 2023; Maassel et al., 2023). That these returns to child safety were realized when many CPS activities were effectively paused is compelling preliminary evidence that drawing down disruptive and coercive interventions can be safely achieved when paired with broad and robust anti-poverty measures (Arons, 2022).

A population-level emphasis on child safety and family well-being has substantial promise as a route forward for system reform. I concur with Berger and Harden that a public health approach that emphasizes ecological causes of child safety and that broadly prioritizes child safety outcomes across federal programs offers a compelling route forward (Feely et al., 2020). These kinds of ecological approaches have the dual virtues of acknowledging the complex (often nonlinear and dynamic) pathways connecting child maltreatment to a broad set of social and family processes at the same time as they highlight how specific local conditions impact child and family stability and safety.

Structural and institutional racism takes diverse forms across places and is often reconfigured over time. Wulczyn (2023) correctly emphasized that careful attention to discrete outcomes for marginalized groups can often provide better insights than broad inequality metrics. And few people are better positioned to identify these specific processes than members of marginalized communities themselves. Participatory action research designs can provide a useful vehicle both for identifying vectors of child and family harm that administrative data systems might miss, and for improving the accessibility and accountability of policy systems to the publics they serve (e.g., Rise Participatory Action Research Team, 2021).

The equity concerns at the heart of critiques of disproportionate foster care placement of Black and Indigenous children can never be addressed by increasing the rate of intervention for White children and families. Even if agency bias plays no role in decision making, a core moral question remains: is it just to separate a child from their families in the interest of safety if the cause of that unsafety is a joint product of contemporary and historical racism? It could only be so under the most extraordinary circumstances, when all other attempts to remediate the harms to families caused by racism have been exhausted.