Search

Three Data-Driven Approaches to Build Police-Community Relations

Authors: Steve Kuo and Scott Kirby

Publish Date: June 27th, 2022


We have an opportunity to be at the forefront in the shift in American policing. Out of the unprecedented support for change, we have found a lack of solution-orientated approaches to police-community relations.” – Founding Director of the Center for Justice Research1

The public appetite for policy change around policing has grown at an unprecedented rate. A July 2020 study released by Gallup found that 58 percent of Americans agree that policing needs major changes compared with only 6 percent that say no change is needed. Moreover, large majorities support an “increased focus on accountability [and] community relations.”

In the ongoing nationwide discussion about police reform, one aspect is frequently cited - how to access and understand information about the use of force in law enforcement agencies? What if cities could build a data driven solution to assist in predicting and avoiding use of force incidents? At the writing of this article, we are aware that a couple large metropolitan Police Departments are collecting and using data to better understand what is driving the incidents. Several large police departments are beginning to use data to identify patterns in behavior to proactively prevent and provide training or other remediation as needed. Some law enforcement agencies are also using this same data to support hiring decisions.


Policymakers, in turn, are reacting strongly to keep pace with public sentiment. According to the National Coalition of State Legislatures, lawmakers introduced 450 pieces of legislation in 31 states in the 11 weeks after George Floyd’s death to increase scrutiny around policing incidents. Are these legislative decisions being made with a clear understanding of the data and an in-depth analysis of the total actions from the policing agency or simply based on media attention?

The discussion demands a methodical process for identifying the outliers and developing transparent strategies to bring change in the most effective way possible. In an increasingly digital landscape, police departments are beginning to collect enough data to really transform the culture of the criminal justice system through analytics. Data analytics can help police departments better understand what has occurred, what insights into staffing can be revealed and what recommendations through analysis could prevent future issues.

Collecting and Capitalizing on Data

First, let’s define the three levels of data analytics complexity and why they are each different and require support from all aspects of the department.

  • Descriptive Analytics: This type of analytics uses the existing data to describe what occurred much like a rear-view mirror. Being able to visualize events that have just happened might allow the department to take precautionary action or simply better understand the decisions officers have made.

  • Predictive Analytics: This type of analytics, typically built by data scientists and statisticians, uses large historical datasets to present various potential future outcomes, based on policy and environmental changes to inform new types of training and outline the potential impact on the departments operations. Although the visualizations are built by data scientists, the analysis of this requires the senior leaders using the outcomes to both understand the data and its limitations (missing data, etc.) and how it can be used to inform decision making.

  • Prescriptive Analytics: This type of analytics requires investment in sophisticated software that allows machines to sift through enormous quantities of data to speed up analysis and reduce data error, and support making decisions that are typically made by the leadership within the department. Think of this as data-driven decision making.

Next, we can look more tactically at the data. Through improved reporting (including the right tools and infrastructure), feature engineering (using domain knowledge to transform attributes into alternative qualitative fields), and advanced analytics, departments can capitalize on collected data (sample data shown in the table below) to make the best decisions for the agency and the community.

Data Type

Sample Attributes

Individual Characteristics

  • Age

  • Gender

  • Rank

  • Bureau/Division

  • Assignment Unit

  • Length of Service

  • Test Scores

  • Personal City/State

Officer History

  • Complaints

  • Commendations and Feedback from the community

  • Incident History, including Use of Force events

  • Administrative Action

  • Accidents/Injuries

Department Data

  • Arrests (by classification and category)

  • Calls for Service (by classification and category)

Open-Source Data

  • Census data (to include neighborhood income distribution and gentrification tracts)

Let’s explore how to leverage analytics across all three levels of complexity using some of the attributes above, to not only inform decision-making, but also facilitative collective problem-solving throughout the department.


Scenario 1 (Descriptive Analytics)


Situation: A specific division has responded to numerous domestic violence events in the last six months and a series of complaints are logged against this division by individuals with first-hand encounters with its officers.

Objective: Eliminate information silos and improve situational awareness by enabling real-time data flow. Determine if complaints are driven by one person or only a few people (rather than the entire unit).


Future State with Analytics: Through enhanced reporting and geographical analysis of incidents at crime control strategy meetings, command staff are alerted to these incidents and are able to immediately reassign officers and/or evaluate which officers may need training. You can look at officer characteristics and activities (such as tenure on the department, unit longevity, types of calls they respond to, uses of force or any performance indicators, etc.). Then based on that, you can determine the appropriate intervention for the individual (e.g., retraining, transfers, services, etc.).


Scenario 2 (Predictive Analytics)


Situation: The department has consistently collected data around incident history, calls-for-service, arrest, and personnel information (such as tenure and performance) and is looking for a way to develop a more sophisticated system to support officer wellness and prevent negative community interactions.


Objective: Create a proactive methodology that utilizes a variety of attributes to identify interventions such as more robust officer support mechanisms (e.g., mental health services), additional training opportunities, among others.


Future State with Analytics: Robust machine learning models (e.g., decision trees and neural networks) are created through the department’s own data and supplemented by complementary data, and the department can identify any emerging behaviors or trends that may warrant intervention.


Scenario 3 (Prescriptive Analytics)


Situation: Attrition is high, and the department needs to maximize its current workforce without overburdening its officers with untested or unnecessary work.


Objective: Evaluate historical trends, forecast the load (e.g., hours worked) on officers, and generate demand-based schedules using mathematical optimization.


Future State with Analytics: Produce an “optimized” schedule that evaluates all available data but is flexible enough to cater to unforeseen changes in staffing. Demonstrate that officer well-being is a priority even when ensuring staffing demands are met.


Conclusion


An agency taking on this type of data analysis requires commitment from all levels in law enforcement. There is certainly a need for financial investment in technology and human resources. Additionally, leaders must invest the time in their own data literacy to trust the information presented within the department.


Public safety is constantly changing. It is essential that public safety leaders identify data driven strategies and tools to guide their agency. More importantly, leaders need to adopt a mindset change with respect to data. Data is not the enemy of good policing, but a way to tell a fuller story to the public on all the ways police keep their communities safe. Moreover, data can be the way to identify the potential challenges within the ranks before they become a public risk and build a better relationship between our police and communities.