Methods: The symposium covers three core problems. The first paper addresses measurement. To understand whether policy and practice changes influence disparity, it is important to ask whether our usual measures of disparity support causal inference when using observational data. The first paper explores (1) the strengths/limitations of our conventional measurement approaches and (2) proposes an approach to resolving the identified issues. The second paper draws on the insights from the first and examine placement rate differences in two ways. Using advanced statistical models drawn from epidemiology, we show how our proposed measurement strategy reveals important insights that link placement, poverty, socio-ecological diversity, and racial disparity cross-sectionally and longitudinally. The third paper uses the Empirical Bayes residual to target areas where placement rate disparities are particularly acute.
Results: The first paper shows why conventional measures of placement rate differences are misaligned with the requirements of causal inference. An alternative approach is proposed. Using the methods proposed, the second study considers Black and White child poverty alongside Black and White child placement from both cross-sectional and longitudinal perspectives. Although Black child placement rates are higher on average, the relationship between poverty rates and placement rates depends on race, a finding that runs counter to one popular narrative. In the third paper, counties are sorted into categories differentiated by how the placement rate for Black and White children deviates from the overall average. In some counties, the county placement rate is above the statewide average but the Black child placement rate is below that average. In other counties, the opposite is true: the county placement rate is below the statewide average but the Black child placement rate is above that average.
Conclusions and Implications: Our findings show that some measures of Black and White child placement rate differences more clearly align with the demands of causal inference than others. When robust measures are used to link poverty and socio-ecological diversity with placement disparity, we find a more nuanced, actionable set of relationships. When we apply the lessons learned to the analysis of county-level placement disparity, it is easier to see where investments in anti-disparity programs are more likely to move the equity needle forward.