Comparing Possible Predictors of Outcome in Police Use of Force Incidents
Faculty Information
Ariana Mendible | Course: DATA 3310 | Completed project
Presentation Type
Individual
Presentation Format
Oral presentation
Start Date
10-5-2024 2:00 PM
End Date
10-5-2024 3:00 PM
Abstract or artist statement
Across the country, social justice advocates and federal investigators alike report unconstitutional and excessive police use of force (UOF) that particularly targets minoritized communities. While some thorough investigations into individual municipalities have been made, certain issues unique to UOF data create barriers to conducting large-scale analysis. Notably, there is no standardized national system of data acquisition and treatment; how UOF incidents are recorded and investigated vary substantially between cities. Though not amending the problem of how data is collected by police, this project explores analysis tools for comparing possible predictors of outcome in police use of force incidents, or what level of force is used. Publicly available UOF datasets are utilized to compare three cities of different population sizes: New Orleans, Seattle, and Chicago. Census data from these cities is also used to provide additional demographic information. Preliminary results show that race may be the statistically strongest indicator of what level of force police use in a given incident, and that Black residents are overrepresented in UOF incidents overall. While future work is needed to uncover, address, and prevent the causes of such patterns, finding ways to recognize and analyze them in the data can be helpful in the initial steps of deconstructing systemic racism and violence.
Keywords: police data, use of force data, machine learning, regression analysis
Comparing Possible Predictors of Outcome in Police Use of Force Incidents
Across the country, social justice advocates and federal investigators alike report unconstitutional and excessive police use of force (UOF) that particularly targets minoritized communities. While some thorough investigations into individual municipalities have been made, certain issues unique to UOF data create barriers to conducting large-scale analysis. Notably, there is no standardized national system of data acquisition and treatment; how UOF incidents are recorded and investigated vary substantially between cities. Though not amending the problem of how data is collected by police, this project explores analysis tools for comparing possible predictors of outcome in police use of force incidents, or what level of force is used. Publicly available UOF datasets are utilized to compare three cities of different population sizes: New Orleans, Seattle, and Chicago. Census data from these cities is also used to provide additional demographic information. Preliminary results show that race may be the statistically strongest indicator of what level of force police use in a given incident, and that Black residents are overrepresented in UOF incidents overall. While future work is needed to uncover, address, and prevent the causes of such patterns, finding ways to recognize and analyze them in the data can be helpful in the initial steps of deconstructing systemic racism and violence.
Keywords: police data, use of force data, machine learning, regression analysis