Evidence of Police Discrimination from a Conjoint Analysis of Police Officers

• In Event: Experiments in South Asian Politics

Abstract

When choosing what cases to investigate, do the police discriminate? We employ a forced conjoint design that holds the guilt of offenders constant to evaluate biases in police officers' preferences for investigation based upon perpetrator attributes. Conducting a regionally representative survey of 1,065 law enforcement officers of all ranks in five districts of Bagmati Pradesh in Nepal, we find evidence of discriminatory investigation practices. Absent constraining protocols that reduce officer discretion, police officers are more likely to target offenders who are from caste-class subjugated communities. In addition, police officers' assessments of institutional investigatory preferences reveal caste-based considerations: police officers believe the police in general are more likely to investigate low-caste offenders than high-caste offenders. These findings add to the body of evidence on whether police discriminate, which has previously focused on use of lethal force and police stops, and further demonstrate that concerns over systemic bias in policing are warranted.

Authors

• Margaret Boittin, Stanford University

• Rachel Fisher, UC Berkeley

• Cecilia Hyunjung Mo, UC Berkeley