Critical Race Theory for Statisticians: incorporating CRT into statistical analyses

Previous analyses of Stand Your Ground (SYG) cases have been primarily descriptive. We examine the relationship between race of the victim and conviction of the defendant in SYG
cases in Florida from 2005 to 2013. We frame our study using Public Health Critical Race Theory Methodology. Data from the Tampa Bay Times SYG database was supplemented with
available online court documents and/or news reports (N=204 cases). The outcome is whether the case resulted in a conviction; covariates include race of the victim (White, non-White),
whether the defendant could have retreated from the situation, whether the defendant pursued the victim, if the victim was unarmed, and who was the initiator of the confrontation. We find
race of the victim to be a significant predictor of case outcome. After controlling for other variables, the defendant is two times (OR = 2.1, 95% CI [1.07, 4.10]) more likely to be convicted
in a case that involve White victims compared to those involving non-White victims. SYG legislation in Florida has a quantifiable racial bias that reveals a leniency in convictions if the
victim is non-White, which provides evidence towards unequal treatment under the law.  

In the world of statistics, we often use race as a covariate or a component of a more sophisticated analysis that has little to do with race. This line of thinking translates into our teaching of statistical analyses in ways that sometimes reinforce systems of racism and disenfranchisement. Thus, in this presentation I will present ways of utilizing a CRT framework in statistical course development and implementation.  

Although the mere mention of critical race theory (CRT) can elicit strong feelings and has become one of the most divisive concepts in the United States, a recent University of Southern California study estimated that nearly 50% of indicated not knowing much about CRT or had never heard of the term at all. With this as the backdrop, many have called into question whether CRT can be measured quantitatively. Similarly, a recent commentary Robinson, Renson and Naimi (2020) argued that persons conducting machine learning analyses in settings involving human health should be literate in structural racism and further assert that this skill is essential in any research involving human health. Here, I present a review of recent work to demonstrate how CRT has been operationalized in various scientific contexts, including illustrating how different terms, constructs and concepts have been used interchangeably in the literature and in other outlets. Additionally, I will present principles for thoughtfully characterizing people and how we can approach thoughtfully representing them more holistically when interpreting statistical models. 

Over the last few decades, policy making to address human trafficking has evolved to be based on more concrete evidence and information. The fields of data science and statistics research have expanded access, collection, and analysis of enormous pools of data. Policy makers, advocates, and administrators have the ability to plan and implement policy around specific needs and populations. Data scientists and statistical researchers can provide clarity and understanding of policy needs within a population or target area, assess whether data-driven results fit the legislative setting to be addressed and become the intersection between the data and policy-makers. Policy makers, government and non-profit administrators, law enforcement, social services and advocates all use information generated by data scientists and statisticians to make decisions every day. This presentation will elaborate on ways that data and research are used in policy formation and highlight examples of successful use of this data in administration and legislation, as well as failures in policy formation, and opportunities for further progress as a data scientist or statistical researcher.