Crisis Control through Administrative Records: Systematic Inconsistencies in Police Data about Stopped Motorists’ Races, 2003-2025

• In Event: Racial and Ethnic Minorities Section Roundtables Session
  In Refereed Roundtable (90 minute): Table Nine: Race and Racism as Structured within and Culturally Propagated through the Criminal Justice System

Abstract

This paper studies how police officers document stopped motorists’ races to illustrate both the complexity of racial classification and policing’s self-interest in legitimacy. Matching 735,166 traffic stops from Florida with drivers’ voter records via probabilistic linkage, I compare officer- documented with drivers’ self‐identified races across groups. Mismatches occur most prominently among drivers self-identifying as Asian and Hispanic for these groups’ permeable and blurry boundaries. In contrast, White and Black motorists are documented mostly the same as how they self-perceive. However, the ambiguity of racial classification alone cannot explain all discrepancies. Despite typically consistent classification, a spike in recording self-identified Black motorists as White appeared in early 2012 and lasted until 2014. Tracing this puzzle through abductive reasoning, I leverage the February 2012 high-profile killing of Trayvon Martin in Florida for its unique relevance. Results from an interrupted time series analysis (n = 22,606 stops) confirm a large post-scandal increase in misclassification, significant only for self-identified Black people and in the direction favorable for law enforcement. I argue that in police race data, strategic anomalies can be hidden within a broader set of seemingly inevitable ones. This insight portrays the challenges of both the task (i.e., racial classification being differentially non-intuitive) and the project of state legibility (i.e., legitimacy being part of the institutional incentive), wherein the latter can co-opt the former and shape administrative data production.

Author

• Khoi Ngo, Washington University in St. Louis