Co-authored by Cassidy Carroll of The Collins Law Firm, P.C.
Can computer algorithms use officers’ behavior to predict police misconduct?
It may sound like something out of a science-fiction movie, but researchers, part of the White House’s Police Data Initiative, are working at the University of Chicago to try to build a better early warning system that predicts when officers will commit police misconduct against civilians. Using over a decade of data from the Charlotte Mecklenburg Police Department, researchers say their algorithm-based early warning system can predict misconduct between police officers and civilians and suggest preventive measures by looking for police misconduct warning signs in the data. The algorithm-based early warning system’s sensitivity and accuracy holds great promise for serving as a method for preventing police misconduct. However, algorithm-based early warning systems to predict police misconduct have been tested before and often met with opposition.
Over 20 years ago, using algorithms to predict police misconduct was uncommon and most of the early warning systems relied on supervisor observations or an officer’s number of complaints. At the time, the Chicago Police Department attempted to create an algorithm-based early warning system to predict police misconduct. Initially, the Fraternal Order of Police, the officers’ union, wanted an algorithm-based early warning system because they felt that it would be more objective and accurate than the previous early warning system. Designed by California Scientific, the algorithm-based early warning system used factors such as officers’ complaint history and stressful personal life events (i.e. recently divorces, serious financial debt) in order to predict which officers engage in police misconduct. After two years, the algorithm-based early warning system was formally shut down because the Fraternal Order of Police objected to administrators using the predictions to preemptively interrogate and recommend counseling.
Today, while the implementation of early warning systems is still being met with some resistance, the use of advanced algorithms to help analyze and predict crime in police departments nationwide is increasing. Instead of identifying police misconduct after the fact, new algorithm-based early warning systems aim to prevent behaviors that lead to police misconduct. In addition to the using complaint history, the University of Chicago group discovered new and more subtle factors that could predict police misconduct, such as officers participating in “stressful” incidents (i.e. suicide, domestic violence). The previous early warning system overestimated the number of officers at risk of engaging police misconduct by 6%. However, the new algorithm-based early warning system flags 51% fewer officers, yet it correctly identified more officers who would engage police misconduct in the next year, suggesting increased accuracy.
Since the algorithm-based early warning system is still being tested, researchers do not know whether their predictions will lead to actions that reduce police misconduct. Therefore, citizens should be on the alert and take the following steps when reporting police misconduct:
Police misconduct can have a significant impact on a person’s health, families, and communities and may leave them fearful of the justice system. If you believe that you or your loved ones may have been the victim of police misconduct, please contact us to learn more about your legal rights.
Sources:
FiveThirtyEight – “We Now Have Algorithms to Predict Police Misconduct” (3/9/16)
"*" indicates required fields