Differential Racial/Ethnic Predictive Validity

Abstract

Recent findings indicate that including White offenders in the sample biases the predictability of risk and needs assessment instruments. As a result, this study examines the predictability of the Los Angeles County Needs Assessment Instrument (LAC) on a sample of African American and Hispanic juvenile probationers. Given that the extant literature focuses on regression analysis, to the curtailment of error analysis, this study also provides a unique examination of predictive error. The results suggest that the instrument under examination predicts better for Hispanics than African Americans. Of the two minority groups, the needs assessment instrument demonstrated the greatest effect size for Hispanic probationers. The LAC performed 16% better than chance predictions when classifying Hispanic juveniles. The area under the curve value was nonsignificant for African American juvenile probationers. The situating of our research findings, their limitations, suggestions for future research, and policy implications are discussed.Keywords predictive validity, juvenile probation, needs assessment, predictive error, race/ethnicity Despite widely accepted racial/ethnic differences within the United States economic, health, educational, and criminal justice systems, minorities are continuously compared to members of the dominant White majority group in offender assessment validations. In a recent review of forensic risk assessment meta-analyses, Singh and Fazel (2010) concluded that, in cross-racial/ethnic assessment comparisons, the inclusion of White offenders in a sample increases the effect size of the predictive instrument under question, ultimately limiting the ability of the instrument to predict accurately for minority groups. Consistent

Racist Algorithms or Systemic Problems? Risk Assessments and Racial Disparities

“…Other studies have defined bias as a difference in error classification or false positive rates, meaning a risk instrument may falsely classify one group as high risk at a higher rate than another group, which could easily result in unfair, harsher treatment of the group with more misclassifications. This approach was taken by a few non-peer-reviewed studies that reported a bias against Black adults (e.g., Angwin et al, 2016; Lason et al, 2016 [both were discredited, see Flores et al, 2016]), and a few peer-reviewed studies since 2000 (e.g., Muir et al, 2020; Rembert et al, 2014; Skeem & Lowenkamp, 2020), which found differences in false positive rates were small to nonsignificant or favored the particular group of color (meaning more individuals who were not categorized as high-risk recidivated), with one exception (see Dressel & Farid, 2018).…”

Unjust Disparities? The Impact of Race on Juvenile Risk Assessment Outcomes

“…Rembert, Henderson, and Pirtle (2014) focused on the predictive error differences across 480 Black and Hispanic/Latino males utilizing the Los Angeles County Needs Assessment Instrument (LAC). Results indicate that overclassification errors were similar for the two groups across three different cutoff scores.…”

On the Precipice of Intersectionality

“…On one hand, numerous findings have demonstrated that of any racial/ethnic group, Black offenders had the greatest likelihood of being classified as high risk (Eisenburg, Bryle, & Fabelo, 2009;Henderson, 2006;Henderson, Daniel, Adams, & Rembert, 2007;Whiteacre, 2006;Yacus, 1998). Research has also demonstrated predictive bias with minority offenders being more likely to be overclassified (i.e., Type I error or false positive) (Rembert et al, 2014;Whiteacre, 2006). Findings additionally indicated a direct relationship between the number of Whites in the sample and the instruments' predictability (Edens, Campbell, & Weir, 2007;Gendreau, Goggin, & Little, 1996;Leistico, Salekin, DeCoster, & Rogers, 2008).…”