Race and Risk

Yesterday the Statesman reported on the continuing saga of "The seven [capital murder] cases identified by [then Attorney General] Cornyn [that] were all tainted by testimony by psychologist Walter Quijano, who regularly told juries that defendants were more likely to commit future criminal acts because they were black or Hispanic."  I remember well when this saga was much younger, I was with TDCJ, and I hosted a group of Argentinian students in a discussion of the American death penalty and Victor Saldano, on death row in Texas from Argentina and one of the seven cases. I showed them his page on the death row website and they said something to effect of "why do you even list his race? You Americans are totally hung up on race." That really stayed with me. 

And yes, we are indeed, but not without good cause at this point in our history, as Quijano's fairly recent testimony suggests.  I am among those who are hung up on racial disproportionality in criminal justice and all of the systems we design to intervene in people's lives from one angle or another. I find it ironic that we have realized it is illegitimate to use race as a proxy for risk in imposing the death penalty, but there is a raging debate in child welfare about whether race is legitimately correlated with higher risk so that we should actually expect disproportionality, or whether racisim in the system itself has a role that we need to be addressing.  My friend Donald Bauman, now retired from DFPS, recently provided me with the following abstract of some cutting-edge research he has conducted:

Findings indicate an important interaction between race, income, and risk of maltreatment. First, we found that poverty is associated with higher risk assessment scores.  We also found that African American families involved in both substantiated and unsubstantiated cases were assessed by caseworkers as having lower risk than White families.  Finally, when controlling for poverty, race is not a significant predictor of substantiated maltreatment.  However, when controlling for risk, poverty was not a significant predictor of substantiation, while race did emerge as a significant predictor.  This suggests that although income may influence risk assessment, it is not a factor that influences the threshold for the substantiation decision.  Rather, the findings suggest that there are racial differences in the risk threshold used by caseworkers in making the substantiation decision.  Specifically, the risk threshold for substantiation is higher for Whites than it is for African Americans.  One possible explanation for this (as also suggested by Rivaux et al., 2008 regarding the decision to place children in care) is the fundamental attribution error (Ross, 1977), a concept from social psychology that refers to the tendency to undervalue situational explanations for the observed behaviors of others and to overvalue personal explanations, such as traits and attitudes.  In the context of the substantiation decision, this would suggest that poverty (a situational factor) is underestimated in favor of race (a personal factor).  Thus, although poverty may be an important factor to address when assessing risk, caseworkers may be assessing the risk associated with poverty differentially for poor African American families than for poor White families.

In other words, race matters in that initial decision to intervene in the child abuse and neglect system.  In the other camp - though we need to find a way to bring these two views harmoniously together - is the work of Elizabeth Bartholet of Harvard, summarized this June by Chapin Hall at the University of Chicago.

This debate is a huge challenge faced by those of us who work to reduce disproportionality by facing the issue of race and racism head-on.

We believe that the evidence presented at this conference signals that it is time for reconsideration of certain past assumptions and conclusions. It indicates that generally there is a significant black/white maltreatment gap, one that roughly parallels the gap in official maltreatment reports. This evidence contradicts the belief that black children are included at high rates in the child welfare system because of racial bias. This is not to say that the evidence presented removes the possibility of bias. Bias may well exist in pockets of the system, operating in ways that lead black children to be either over- or underserved, and it is present more generally within the larger society. But we find no evidence that initiatives that emphasize reducing the high representation of black children will provide a path to more equitable services.