State law now requires every law enforcement agency to implement an “early warning system.” What is an early warning system? Do such systems work? And what can small agencies do to comply with the law? Read on to learn more.
The basics. S.L. 2021-138, more commonly known as SB300, enacted new G.S. 17A-10, which requires every agency that employs certified officers or deputies to “develop and implement an early warning system” to track and improve officers’ performance. Here’s the pertinent text in full:
PART VIII. DEVELOPMENT OF EARLY WARNING SYSTEMS
SECTION 8.(a) Chapter 17A of the General Statutes is amended by adding a new section to read:
17A-10. Development of law enforcement early warning system.
(a) Every agency in the State that employs personnel certified by the North Carolina Criminal Justice Education and Training Standards Commission or the North Carolina Sheriffs’ Education and Training Standards Commission shall develop and implement an early warning system to document and track the actions and behaviors of law enforcement officers for the purpose of intervening and improving performance. The early warning system required by this section shall include information, at a minimum, regarding the following:
(1) Instances of the discharge of a firearm.
(2) Instances of use of force.
(3) Vehicle collisions.
(4) Citizen complaints.
(b) Information collected under this section that is confidential under State or federal law shall remain confidential.
(c) For purposes of this section, “law enforcement officer” means any sworn law enforcement officers with the power of arrest, both State and local.”
SECTION 8.(b) This section becomes effective December 1, 2021, and applies to actions and behaviors on or after that date.
Some provisions of the bill likely need to be interpreted with some common sense. For example, the requirement that agencies track “[i]ncidents of the discharge of a firearm” should not be read to mean that every time an officer goes to the firing range to qualify with his or her service weapon, that needs to be recorded in the early warning system.
What is an early warning system? An early warning system, or EWS, is a system designed to identify officers who may be at risk of using excessive force or otherwise acting improperly with members of the public. Such systems look at factors such as the quantity of complaints an officer has received, the number of times an officer has used force, or the number of times an officer has engaged in a high-speed vehicle pursuit. If an officer is identified by the system, the agency may review the officer’s performance and, if appropriate, may provide additional training or other support to the officer. Some prefer to call these systems early intervention systems, or EIS, to emphasize the importance of taking action to change an officer’s behavior once the officer has been identified by the system. These systems are normally non-disciplinary, meaning that agencies do not impose any formal sanctions on an officer as a result of the officer being flagged by the system. Of course, some officers view being identified by an early warning system as a likely impediment to promotion.
How long have they been around? The first early intervention systems were developed in the 1970s, when research began to support the idea that a relatively small number of officers are responsible for a disproportionate share of citizen complaints and uses of excessive force. Some agencies began to attempt to identify those officers, each creating its own system for doing so. In 1981, the U.S. Commission on Civil Rights issued a report called Who Is Guarding the Guardians? A Report on Police Practices, which suggested that early warning systems could help identify “violence-prone officers.” Id. at 81. The Commission appears to have based its finding on promising early experiences in a handful of large departments, including Oakland, New York City, and Kansas City.
How widespread are they? Most large departments today have some sort of early intervention system, whether built internally or purchased off-the-shelf from one of the several software vendors that operate in this space. A recent survey revealed that “in 2016, 69.3% of the . . . police departments in the U.S. with 100 or more officers had an EIS, and 80.5% of the . . . agencies with 500 or more officers had such a system in place.” Council on Criminal Justice, Policy Assessment: Early Intervention Systems (May 2021). CALEA accreditation requires the implementation of an EIS. As discussed further below, smaller agencies may be less likely to have such systems, and the systems that they do have may be less sophisticated.
How do they work? What factors do they consider? In their simplest form, early intervention systems track certain events, like uses of force, discharges of a firearm, lawsuits against an officer, or complaints. Agencies vary in the types of events they track, and a recent primer by the Police Foundation identified seventeen different types of events that may be part of a system, including an officer’s failure to appear in court, excessive use of sick leave, and the frequency of warrantless searches and seizures. Karen L. Amendola & Robert C. Davis, Best Practices in Early Intervention System Implementation and Use in Law Enforcement Agencies, National Police Foundation (Nov. 2018). If an officer has more than a specified number of incidents over a defined period of time – for example, more than two complaints over a two-year period – the officer is flagged for review. The review may show no problem. An officer may have three complaints that all arose from a single incident in which the department determined that the officer acted correctly. Or the review may show that the officer could benefit from additional training or other support, which the department would then provide.
Some agencies have moved away from simple event-threshold early intervention systems and towards systems based on “predictive analytics” or “machine learning.” As I understand the latter type of system, the system does not look at a defined set of events using fixed thresholds. Rather, it constantly analyzes which officers have been involved in adverse events and looks at those officers’ backgrounds to identify commonalities. Based on the system’s ever-evolving understanding, it flags officers who appear to be at an elevated risk of being involved in future adverse events. Ideally, such a system would adjust for an officer’s duty assignment, which may impact the likelihood of certain events taking place. There is some evidence that these sophisticated systems may be more accurate. See Jennifer Helsby et al., Early Intervention Systems: Predicting Adverse Interactions Between Police and the Public, 29 Criminal Justice Policy Review 190, 190-209 (2017) (claiming that the authors “have developed a data-driven EIS that uses a diverse set of data sources from the Charlotte-Mecklenburg Police Department and machine learning techniques to more accurately predict the officers who will have an adverse event. Our approach is able to significantly improve accuracy compared with their existing EIS: Preliminary results indicate a 20% reduction in false positives and a 75% increase in true positives.”).
Are they effective? There is limited research on this question, and the results are somewhat mixed. A recent review of the social science literature concluded that “empirical studies have generally shown positive results,” in terms of “a reduction of complaints, uses of force or both,” while a smaller number of studies “have shown mixed results or no effect on officer behavior or performance,” or even “a decline in officer productivity,” perhaps as officers pull back from enforcement activity in an effort to avoid interactions that could lead to being flagged by the system. In short, “more research is needed.” Christi L. Gullion & William R. King, Early Intervention Systems for Police: A State-of-the-Art Review, 43 Policing 643, 643-58 (2020). A 2021 research summary from the University of Chicago likewise characterized the literature on the effectiveness of these systems as “extremely limited.” Zoe Russek & Dylan Fitzpatrick, Early Intervention Systems, University of Chicago Urban Labs/Crime Lab (2021). North Carolina’s new mandate may provide a natural before-and-after experiment for social scientists to analyze.
How can small agencies implement them? Some may argue that small agencies have less need for early warning systems because supervisors have more direct knowledge of potential problem officers. In other words, it may be harder to fly under the radar in a small department. Whatever the merits of that argument, the General Assembly did not exempt small agencies from the early intervention system mandate. There are hundreds of small agencies in North Carolina, many with limited technological and financial resources. SB300 does not allocate any funds or provide any technical assistance to assist these departments in establishing early intervention systems.
Small agencies may be able to comply with the law by tracking the statutorily-mandated events manually on a spreadsheet, reviewing the spreadsheet monthly or quarterly to determine whether troubling patterns exist. The efficacy of such a system is an open question, and ensuring that the relevant events are consistently reported and included on the spreadsheet may require a significant organizational commitment. I would be interested in hearing from small agencies that have successfully implemented an early intervention system – please reach out to me directly by email or phone.
What if an agency fails to implement an EIS? SB300 does not provide for a specific consequence for agencies that do not establish a system. However, an agency that fails to create such a system, or that implements an ineffective system, may face increased liability risk, especially in light of the mandate. See, e.g., Wilson v. Miami-Dade County, 370 F.Supp.2d 1250 (S.D. Fla. 2005) (allowing a wrongful death lawsuit to proceed that was based in part on a police department’s failure to implement an effective early warning system); Estate of Heenan ex rel. Heenan v. City of Madison, 111 F.Supp.3d 929 (W.D. Wisc., June 1, 2015) (in a wrongful death case, denying a city’s motion for summary judgment regarding Monell liability because “[t]he evidence plaintiff has proffered on summary judgment . . . permit[s] a finding by a reasonable trier of fact that the City acted with deliberate indifference in failing to conduct meaningful investigations of civilian complaints of excessive force generally, and to ensure that supervisors were aware of repeated complaints against a single officer . . . by means of an early warning system or other mechanism for monitoring trends in civilian complaints”).
Additional resources. Those interested in further reading may wish to explore the links in this post or to review this document prepared by the North Carolina Department of Justice. The NCDOJ document provides an example of a simple spreadsheet-based EIS intended to comply with SB300, and it links to several North Carolina agencies’ EIS policies.