An exploration of recurring early intervention (EI) alerts to address at-risk officers through the lens of police reform

Abstract

Police reform, including transparency and officer and agency accountability, is a top priority. In response to patterns of excessive force and misconduct, reform agreements have been instituted, and consistently require early intervention (EI) systems, a supervision and accountability tool to address at-risk officers. Yet little is known about the likelihood and timing of recurring EI alerts. This study examines EI data from a large, metropolitan police agency in the southwestern USA. Results indicate officer race, division, time to initial alert, type of performance indicator, and year of the alert were significant for the likelihood and timing of subsequent EI alerts. Supervisor tenure was also significant for timing to a subsequent EI alert. For officers with a subsequent EI alert, half recurred within 6 months of their initial EI alert and intervention, and 90 per cent did so within 2 years. Implications for police reform include enhanced supervision and accountability with EI systems.

Police agencies are under continual pressure to address local concerns in the communities they serve and national issues of police reform. Since the early 1990s, the U.S. Department of Justice (DOJ) has attempted to address police reform through federal oversight, when formal investigations of police agencies find patterns of excessive force or misconduct (U.S. DOJ Civil Rights Division 2017; Gullion and King 2020; Archbold 2021). Police reform considers how to improve officer performance, supervision, and accountability, frequently recommending early intervention (EI) systems. EI systems track data to identify at-risk officers early and prevent recurring problematic behavior and misconduct through the review and execution of EI alerts and interventions. In addition, EI systems have also been recommended as an essential accountability tool by police leadership and experts for the past five decades (Walker and Archbold 2013; U.S. DOJ Community Oriented Policing Services [COPS] Office 2019). Not surprisingly, almost 70 per cent of US police agencies with 100 or more officers had an EI system as of mid-2016 (U.S. DOJ Bureau of Justice Statistics [BJS] Law Enforcement Management and Administrative Statistics Survey [LEMAS] 2020).

Prior EI research has primarily focused on the change in officer activities pre–post EI alert, finding mostly positive reductions in complaints, use of force, and other activities (Walker et al. 2001; Davis et al. 2002; Lersch et al. 2006; Macintyre et al. 2008; Bobb et al. 2009; Worden et al. 2013; Shjarback 2015; Briody and Prenzler 2020). Other studies have explored the most predictive performance indicators, finding that a prior history of at-risk behavior and misconduct strongly predicted future misconduct (Carton et al. 2016; Helsby et al. 2018; Cubitt et al. 2020). Limited research has also explored the likelihood of officer activities and EI alerts, finding that officers with multiple annual complaints were at a higher risk of being terminated for misconduct (Kane and White 2009), and more excessive force complaints increased officers’ likelihood of using higher levels of force (McCluskey and Terrill 2005). For predictors of EI alerts, Gullion and Ingram (2024) found that officer tenure, division, supervisor gender, type of performance indicator, number of prior EI alerts, and year of the EI alert were significant for the likelihood of a subsequent EI alert. Regarding the timing of officer activities, studies find that the early onset of officer misconduct including complaints and uses of force increases the risk of a longer duration and higher frequency of problematic behavior (Kane and White 2009; Harris 2014; Harris and Worden 2014; Gullion et al. 2021, 2023).

While limited research has examined the likelihood and timing of repeated officer behavior and the likelihood of EI alerts, the timing between recurring EI alerts has not yet been examined. This study hopes to contribute to this gap by examining the predictors impacting the likelihood and timing between an officer’s initial and subsequent EI alert using longitudinal data from a large metropolitan police agency in the southwestern USA. This examination will provide a more complete picture of repeated at-risk officer behavior to improve the effectiveness of EI systems including the review and execution of EI alerts and interventions with officers.

POLICE REFORM

To sustain police legitimacy and community trust, ensuring agency and officer accountability is vital. Yet concerns surrounding high-profile and critical incidents of excessive force and officer misconduct have led to earnest discussions among police agencies and community members about how to improve transparency and accountability in policing. Between 1994 and 2023, patterns or practices of excessive force and officer misconduct have resulted in seventy-nine formal investigations and forty court-ordered reform agreements (federal court-enforced consent decrees and settlement agreements or memorandum of agreements) with various police agencies to bring about necessary change (U.S. DOJ Civil Rights Division 2017; National Policing Institute n.d.). Reform agreements have discussed improving systems for supervising officers and holding them accountable and using data about police activity to identify and correct patterns of police misconduct (U.S. DOJ Civil Rights Division 2017). Police reform required under federal oversight by the U.S. DOJ Civil Rights Division include internal affairs systems (complaint and use of force investigations and discipline), de-escalation, body-worn cameras, community policing, civilian review boards, internal audits, and EI systems, among other topics (U.S. DOJ Civil Rights Division 2017; Archbold 2021; National Policing Institute n.d.).

EI SYSTEMS AND POLICE REFORM

While police reform can include a multitude of topic areas to bring about necessary change, within the seventy-nine formal investigations and forty police reform court-ordered agreements entered by the U.S. DOJ Civil Rights Division (2017) with various police agencies between 1994 and 2023, EI systems have been a consistent requirement. Specifically, EI systems were a required area of police reform in thirty-one of the fifty-five federal interventions (56.36 per cent) and were the seventh most frequently required area among forty-two different areas of police reform (National Policing Institute, n.d.). These reform agreements ‘emphasize not only the creation of [EI] systems, but also the requirement that police leadership and supervisors analyze the data gathered by these systems, address emerging patterns of police misconduct, and enhance individual officer accountability’ (U.S. DOJ Civil Rights Division 2017: 31).

Beyond these federal court-enforced consent decrees and settlement or memorandum agreements, other policing experts and entities arguing for police reform have also consistently included EI systems in their recommendations (Walker and Archbold 2013; U.S. DOJ COPS Office 2019; Gullion and King 2020; Archbold 2021). In terms of importance, the U.S. DOJ COPS Office (2019) argues that EI systems can be a valuable supervisory management tool to increase agency accountability and improve performance, offering a more holistic view of officer behavior. Adding, ‘an EI program is meant to be non-disciplinary, [identifying] employees in need of assistance early on, enabling [agencies] to intervene with the appropriate support to prevent a future incident’ (COPS Office 2019: 61). Furthermore, a Police Executive Research Forum (PERF) survey (2020) of 375 police executives reported that addressing the call for police reform is police leaders’ second most important issue in policing for 2021 and beyond, and that EI systems were one of their top ten requests for federal funding. Thus, police reform and EI systems continue to be a top priority and interrelated goal for oversight bodies, police leadership, and experts throughout the USA.

EI SYSTEMS DESIGN AND PURPOSE

As a non-disciplinary, supervisory accountability tool, an EI system tracks performance indicators (i.e. officer behaviors) and triggers alerts when an officer reaches or exceeds an agency-defined threshold (e.g. two uses of force in 2 months, two complaints in 3 months, etc.) or combination of thresholds (e.g. three of any performance indicators in 4 months). Then a supervisory review of the specific incidents and at-risk behavior(s) identified by the EI system and any relevant officer work history is conducted. The supervisor may meet or speak with the officer to discuss the EI alert and conduct an intervention, and then determine if any formal action is needed. Such interventions may be documented depending on the agency’s policy and practices and follow-up may include post-intervention monitoring to ensure the officer completed the formal action required. Thus, an intervention can be an informal or formal conversation or meeting with the officer, regardless of whether the supervisor determines that formal action (outcome) is needed.

During a supervisor’s review of the EI alert and intervention with the officer, there is also an opportunity to determine if there is a pattern of behavior occurring. If a pattern is found, ensuring the intervention and subsequent formal action (if needed) is tailored and appropriate to that officer is the key to modifying the at-risk behavior (Walker et al. 2001; Macintyre et al. 2008; Gullion et al. 2021, 2023). The effectiveness of an EI system may rely on the identification of at-risk behavior or patterns, whether interventions are tailored to address that officers’ needs, and the execution of interventions, including oversight and accountability by managers. If the review and execution of EI alerts and interventions are ineffective, future adverse events and misconduct are more likely to occur.

PRIOR EI RESEARCH

Research consistently finds that a small percentage of officers account for a disproportionate amount of problematic behavior and high-risk incidents (Alpert and Walker 2000; McElvain and Kposowa 2004, 2008; McCluskey and Terrill 2005; Harris 2009, 2014; Kane and White 2009; Terrill and Ingram 2016). Many police accountability tools exist that can address or prevent problematic behavior, but an effective EI system can do both. While most mid- to large-sized US police agencies have implemented an EI system (U.S. DOJ BJS LEMAS 2020), few EI studies have been conducted.

Most prior EI studies reported positive findings, including that EI systems were effective at identifying and addressing at-risk officers and reducing misconduct for those receiving interventions (Walker et al. 2001; Davis et al. 2002; Lersch et al. 2006; Macintyre et al. 2008; Bazley et al. 2009; Bobb et al. 2009; Worden et al. 2013; Shjarback 2015; Carton et al. 2016; Helsby et al. 2018; Briody and Prenzler 2020; Cubitt et al. 2020; James et al. 2020). Studies found between a 61 and 100 per cent reduction in complaints 1–3 years after the intervention with officers (Walker et al. 2001; Macintyre et al. 2008; Bobb et al. 2009; Briody and Prenzler 2020), decreases in arrests and traffic citations after an intervention (Davis et al. 2002; Worden et al. 2013), or a reduction in force-related lawsuits (Bobb et al. 2009). While some studies found positive reductions in the use of force (Walker et al. 2001; Davis et al. 2002; Bobb et al. 2009), others reported increases in the use of force, and EI systems were not identifying the appropriate at-risk officers (Lersch et al. 2006; Bazley et al. 2009; Worden et al. 2013; James et al. 2020).

Prior research has also explored which performance indicators best predicted at-risk officers and future misconduct (Carton et al. 2016; Helsby et al. 2018; Cubitt et al. 2020). The research found that officers with a prior history of adverse events and misconduct were the most predictive performance indicators of future adverse events (Carton et al. 2016; Helsby et al. 2018) and officers with prior serious misconduct and secondary employment strongly predicted future misconduct (Cubitt et al. 2020). Given this, an effective EI system should appropriately flag such officers for interventions to modify repeated at-risk behavior.

LIKELIHOOD OF AND TIMING BETWEEN RECURRING OFFICER ACTIVITIES

Beyond an EI system identifying at-risk officers, understanding the likelihood and timing between recurring officer activities is key given the opportunity supervisors have to address and prevent recurring at-risk behavior by providing effective EI interventions with officers. First, research has found a high concentration of complaints (Lersch et al. 2006; Terrill and Ingram 2016) and uses of force (McElvain and Kposowa 2004; Brandl and Stroshine 2013) among a small group of officers. For example, 37 per cent of patrol officers across eight US police agencies accounted for all complaints (Terrill and Ingram 2016). Similarly, officers with three or more uses of force accounted for 32 per cent of all uses of force (Brandl and Stroshine 2013), and 24 per cent of officers were the subject of all use of force investigations over 5 years (McElvain and Kposowa 2004).

Few studies have examined the likelihood of recurring officer activities. In a large southwestern US agency, a longer time for an initial complaint and division assignment reduced the likelihood of future complaints (Gullion et al. 2021), while an officer’s initial use of force occurring in neighborhoods with higher minority composition reduced the risk of future use of force (Gullion et al. 2023). Relatedly, a prior history of officer-involved shootings increased officers’ risk of involvement in future shootings (McElvain and Kposowa 2008). Kane and White (2009) found that officers with multiple complaints annually were at an increased risk to be terminated for misconduct, and McCluskey and Terrill (2005) found officers with more excessive force complaints increased their risk of using higher levels of force.

Finally, for the timing between repeated officer activities, research finds officers receive more complaints and uses of force in the first 3 years of their career and are at greater risk for continued misconduct dependent upon the proximity of the first complaint or use of force to the start of their career (Harris 2009, 2014; Gullion et al. 2021, 2023). Gullion et al. (2021) found that for officers who received a subsequent complaint after their initial complaint, more than half did so within the first year, and 94 per cent did so within the first 3 years. Similarly, Gullion et al. (2023) found time to a subsequent use of force was longer when officers’ initial use of force resulted in citizen injury. For officers with a subsequent use of force, more than half did so within 3 months of their initial use of force, and 93 per cent within a year.

LIKELIHOOD OF RECURRING EI ALERTS

Given these findings on the likelihood of and timing between officer activities, insight into the likelihood and timing of recurring EI alerts is crucial to improve EI interventions, modify officer behavior, and ensure the effectiveness of EI systems. Gullion and Ingram (2024) examined the likelihood of EI alerts occurring between 2014 and 2020 in a large southwestern US police agency and found EI alerts occurring in the south division and each additional month of an officer’s tenure decreased the risk of a subsequent EI alert, while EI alerts reviewed by male supervisors increased the risk of a subsequent alert. In addition, chargeable vehicle accidents, unconfirmed sick leave, other performance indicators, as well as a combination threshold, and EI alerts triggered from 2016 to 2020 decreased the risk of a subsequent EI alert. Yet for each additional prior EI alert received, the risk of a subsequent EI alert increased (Gullion and Ingram 2024).

However, prior research has yet to examine the timing between recurring EI alerts. Understanding the risk factors for the likelihood of officer activities and misconduct can help define thresholds which produce EI alerts for review. For example, understanding repeated EI alerts based on officer assignment, the type of performance indicator, or time period of the EI alert can assist in establishing more accurate threshold counts. Also, awareness of the risk factors for the timing of repeated officer activities and misconduct can inform supervisors how to best handle EI interventions with officers to modify at-risk behavior and prevent future misconduct, including ‘how intensive the intervention and remediation should be for individual officers and whether a more comprehensive and prolonged supplementary training or increased supervision would be more appropriate for a group of officers’ (Gullion et al. 2023: 1650). Knowledge of these risk factors including those related to the EI process can also provide insights for managers on whether their supervisors are appropriately handling the review and execution of EI alerts and interventions and how to provide better managerial oversight to hold supervisors accountable.

CURRENT STUDY

Given the paucity of scholarship exploring the likelihood and timing of EI alerts and interventions, this current study fills this critical gap and expands on Gullion and Ingram’s (2024) study of the risk of subsequent EI alerts by focusing on the likelihood and timing between initial and subsequent EI alerts during this study. Using EI data from 2014 to 2020, the goal is to determine whether and to what extent EI alerts and interventions are having the desired impact of modifying at-risk officer behavior. This study explores two research questions:

  

METHOD

Data

Data collected for this study include EI alerts and interventions handled by first-line supervisors that occurred between 2014 and 2020 provided by a large metropolitan police agency in the southwestern USA. Two data sources were drawn from administrative data on EI alerts and employee demographics and supervisor EIS response memos. This EI system tracked nine performance indicators of varying thresholds in their risk management system in 2014. EI alerts trigger if an officer meets or exceeds any agency-defined thresholds (see Table 1).¹

As required by policy, the supervisor reviews the behavior or incidents associated with the EI alert and any other relevant data, meets with the officer to ask about any personal or job-related stressors, and determines an intervention outcome, if any. Subsequently, the supervisor completes a response memo and forwards it through their chain of command. The response memo must summarize each incident associated with the EI alert and articulate the intervention chosen (e.g. coaching/mentoring, referral to services, training, or reassignment) if any, with justification. This meeting with the officer and supervisor response memo is the first step of an EI intervention. Any formal outcome requires monitoring and reporting of the officer’s progress to their division commander. In mid-2017, this agency’s EI process reduced summarizing incidents to a recommendation and asking officers about stressors became a requirement.

Sample

The analysis was limited to officers employed between 1 January 2014 and 31 December 2020, with some officers separated (deceased, resigned, retired, or terminated) from the agency during this time.² Between 2014 and 2020, there were 647 total officers that could have been flagged for an EI alert, and 112 total first-line supervisors (i.e. sergeants) that could have handled the EI interventions with these officers. Data included demographic and occupational information for officers involved in the incidents and details surrounding each EI alert and intervention coded from the supervisor response memos (e.g. type of performance indicator, date of EI alert, date of response memo, EI process details, intervention outcome, etc.). The agency was either not able to locate or could not provide (due to ongoing investigation or litigation) sixteen memos, and an additional ten memos had missing information and were excluded to ensure information from the memos could be matched with administrative data. Thus, the final sample comprised of 256 officers that received at least one EI alert during this study period.

Measures

Dependent variable

The dependent variable was captured in two ways. First, as a dichotomous indicator of whether an officer received a subsequent EI alert after that officer’s initial EI alert and intervention during the study period (1 = yes; 0 = no). Overall, 48.44 per cent of officers received a subsequent EI alert after their initial EI alert and intervention (see Table 2 for full descriptives). The second treatment of the outcome measure was the temporal distance between an initial EI alert and intervention and the first subsequent EI alert (i.e. time to failure), or alternatively, the survival time (in months). This allows for an analysis of the factors associated with time to the subsequent EI alert.³ Of the 124 officers who received a subsequent EI alert after their initial EI alert and intervention (i.e. meeting/speaking with the officer to discuss the incidents that triggered the alert, asking about stressors, and determining whether formal action is needed), the average time was 11 months before failure (SD = 13 months).

Demographic and occupational variables

Demographic and occupational variables associated with the officer’s initial EI alert and intervention were included in the analyses. Officer and supervisor race/ethnicity were both measured through a series of dummy variables, Black, Hispanic, and other (Asian and two or more races) with White as the reference category. Officer and supervisor gender were also measured as dummy variables (0 = female; 1 = male). Most officers (83.20 per cent) and supervisors (84.77 per cent) were male and White (63.67 per cent of officers and 60.94 per cent of supervisors). Officer division assignment was included to control for differences in geographic areas, assigned duties, workload, and supervisors handling of alerts and interventions across divisions, with north division as the reference category (27.73 per cent of EI alerts).⁴ In addition, officer division is important as it is unclear whether the link between officers assigned to high crime areas and higher at-risk incidents such as uses of force is due to officer misconduct or activity level (e.g. Lersch et al. 2006). Supervisor tenure, measured in years, represents the time from their hire date to the supervisor response memo date (M = 15, SD = 6).

EI case variables

EI case variables associated with the officer’s initial EI alert and intervention were included in the analyses. The type of performance indicator was measured through a series of dummy variables: chargeable vehicle accidents; unconfirmed sick leave; other (alleged racial profiling, formal and citizen complaints, missed court; supervisor-initiated discipline; and vehicle pursuits); and a combination of any five or seven performance indicators in 6 months, with use of force as the reference category (37.50 per cent of performance indicators). Time to the initial EI alert represents the time, in months, from the officer’s hire date to their initial EI alert received.⁵ Finally, given the substantive changes to the EI process during this time, the year of the EI alert was measured as a series of dummy variables, 2016 to mid-2017 and mid-2017 to 2020, with 2014 to 2015 as the reference category (50.00 per cent of EI alerts).

EI process variables

EI process variables are the agency’s policy or process requirements for the supervisor’s review and handling of the officer’s initial EI alert and intervention, as detailed in the response memo, and were included in the analyses.⁶ Two EI process variables were captured as dichotomous indicators of whether the supervisor met or spoke with the officer (1 = yes; 0 = no) and provided individual summaries of the incidents triggered by the EI alert (1 = yes; 0 = no). Supervisors met or spoke with officers in 80.47 per cent of EI alerts and provided individual summaries in 50.00 per cent of EI alerts. Finally, supervisors asking officers about any personal or job-related stressors were measured through a series of dummy variables: the supervisor explicitly asked the officer about any personal or job-related stressors; the supervisor implicitly addressed (e.g. determined, discovered, did not find any stressors affecting the officer, did not find any cause for referral, etc.) any personal or job-related stressors with the officer; and the supervisor did not address any personal or job-related stressors with the officer as the reference category (39.45 per cent of the EI alerts).

Analytical strategy

The goal of this study is to examine the factors that affect the likelihood of and timing between EI alerts. First, logistic regression analysis examined the likelihood of officers’ receiving a subsequent EI alert after their initial EI alert and intervention, and factors that may affect this likelihood. Second, survival analysis examined the timing between an officer’s initial and a subsequent EI alert (i.e. time to failure), and the factors that may affect this timing. Survival analysis analyzes time at risk (i.e. time from an officer’s initial EI alert to a subsequent EI alert), while accounting for those who did not experience the event. Officers were right-censored if they did not receive a subsequent EI alert or were separated or terminated from the agency before the study end, 31 December 2020. If so, the end date was the date of their separation/termination.

While this study examined the timing between an officer’s initial and subsequent EI alert occurring between 2014 and 2020, this may not be the first EI alert and intervention the officer ever received. However, the officer’s initial EI alert and intervention during this study is simply a baseline to examine the timing between EI alerts. Regardless of when an EI alert and intervention (i.e. informal/formal meeting to discuss incidents, ask about stressors, and determine outcome) occurs, if appropriately handled by the supervisor, timing to another EI alert should be longer than those handled less effectively. Moreover, given EI thresholds are based on multiple incidents within a certain time, insight into what factors may reduce officers’ risk of a subsequent EI alert is key, regardless of which two EI alerts and interventions are considered.

The second step examines the relationship between one and more factors with the survival time. Cox regression (Cox 1972) was chosen as there is no requirement to identify a particular baseline hazard rate, or starting point, and aligns with approaches by Gullion et al. (2021, 2023), Harris (2014), and Harris and Worden (2014) in their examinations of the likelihood and timing of uses of force and complaints. While prior research finds younger officers tend to make more arrests, use force more frequently, engage in at-risk behavior, and thus, are likely to have more opportunities to receive EI alerts and interventions, there is no specific theoretical or empirical support for a baseline timeframe when this begins (Harris 2014; Harris and Worden 2014).⁷

RESULTS

Logistic regression and survival models were produced to predict the likelihood and timing of subsequent EI alerts. First, a life table clusters the data into intervals for presenting the survival functions, prior to considering the effects of the covariates. The results of the logistic and Cox regression analyses are presented to examine the likelihood and timing of subsequent EI alerts.

Time to a subsequent EI alert distribution

Results of the life table analysis represent an aggregate count of officers who received a subsequent EI alert, clustered into 3-month intervals (see Table 3). This shows the total number of officers at the beginning of the sample, the number of failures or officers who received a subsequent EI alert, and officers censored, or who did not receive a subsequent EI alert during this time. This table also demonstrates the aggregate proportion of officers who survive, which are officers who do not receive a subsequent EI alert at the end of each interval.

Of the 256 officers in the sample, 124 officers or 48.44 per cent failed, receiving a subsequent EI alert before the end of the study. Within 6 months, 58 of the 124 officers, or nearly 50 per cent, received a subsequent EI alert following their initial EI alert and intervention. Finally, 112 of the 124 officers or 90 per cent failed, receiving a subsequent EI alert within the first 2 years of their initial EI alert and intervention.

Related to EI alerts and interventions received early in an officer’s career, 37 of the 124 officers or 29.84 per cent were hired after 1 January 2014, and received their subsequent EI alert within the first 2 years of their career. Specifically, twenty-seven of these thirty-seven officers received their subsequent EI alert with 1 year of the initial EI alert and intervention received in their career, while the remaining ten officers received their subsequent EI alert within the first 2 years of their career.

Likelihood and timing of a subsequent EI alert

Next, results from the logistic and Cox regression analyses examining the likelihood of and timing of a subsequent EI alert after an officer’s initial EI alert and intervention demonstrated that several demographic, occupational, and EI case factors were significant in both models, with few exceptions (see Table 4). In a Cox regression model, positive coefficients indicate a greater rate of risk of failure, or shorter survival times, whereas negative coefficients indicate a slower rate of risk of failure, or longer survival times (Cox 1972).

First, results revealed that being a Hispanic officer was significant in both the logistic and Cox regression models. Thus, Hispanic officers who received their initial EI alert and intervention during this study decreased their likelihood of receiving a subsequent EI alert by a factor of 0.36 and decreased their rate of failure by a factor of 0.40, having a longer time to a subsequent EI alert compared to White officers. In addition, the south, west, and other divisions were significant in the logistic regression model, while only the south division was significant in the Cox regression model. Thus, officers assigned to the south, west, and other divisions when they received their initial EI alert and intervention decreased their likelihood of receiving a subsequent EI alert by a factor of 0.25, 0.41, and 0.35, respectively, while officers assigned to the south division also decreased their rate of failure by a factor of 0.39, having a longer time to a subsequent EI alert compared to officers assigned to the north division. While supervisor tenure was not significant in the logistic regression model, it was significant for timing. Thus, officers who had more tenured supervisors handle their initial EI alert and intervention increased their rate of failure by a factor of 1.04, having a shorter time to a subsequent EI alert, compared to officers who had less tenured supervisors handle their initial EI alert and intervention.

Second, performance indicators for chargeable vehicle accidents, unconfirmed sick leave, as well as a combination of any five or seven indicators in 6 months were significant in both logistic and Cox regression models. Thus, officers who received their initial EI alert based on these performance indicators decreased their likelihood of receiving a subsequent EI alert by a factor of 0.25, 0.34, and 0.32, respectively, and decreased their rate of failure by a factor of 0.40, 0.56, and 0.43, respectively, resulting in a longer time to a subsequent EI alert, compared to officers whose initial EI alert was for use of force.

Additionally, the time to the initial EI alert was also significant in both models. Thus, for each additional month it takes an officer to receive an initial EI alert, the odds of receiving a subsequent EI alert and rate of failure both decreased by a factor of 0.99, resulting in a longer time to a subsequent EI alert.

Finally, EI alerts from 2016 to mid-2017 and mid-2017 to 2020 were significant in both models. Thus, officers whose initial EI alerts were from 2016 to mid-2017 and mid-2017 to 2020 decreased their odds of receiving a subsequent EI alert by a factor of 0.28 and 0.12, respectively, and decreased their rate of failure by a factor of 0.47 and 0.38, respectively, resulting in a longer time to a subsequent EI alert, compared to officers who received their initial EI alert from 2014 to 2015.

DISCUSSION

The purpose of this study was to examine the likelihood and timing of repeated EI alerts. The logistic regression and survival analyses produced similar results with a few exceptions, and several key findings emerged that provide insights for effective EI systems and police reform.

First, the time to an officer’s initial EI alert was significant. Thus, a longer time until the initial EI alert reduced the likelihood of a subsequent EI alert and resulted in a longer time for a subsequent EI alert. In addition, those officers who received a subsequent EI alert and were hired after this study began, received that subsequent EI alert within the first 2 years of their career. This aligns with prior literature that officers with an early onset of misconduct increased their likelihood and shorten their time to repeated misconduct (e.g. Harris 2009, 2014; Harris and Worden 2014; Gullion et al. 2021, 2023). Furthermore, most officers received a subsequent EI alert within a relatively short time. This suggests the timing of EI alerts is key, and supervisors providing more appropriate interventions earlier in an officer’s career have the greatest opportunity to modify at-risk behavior and prevent repeated EI alerts.

Second, Hispanic officers’ risk of receiving a subsequent EI alert decreased and resulted in a longer time for a subsequent EI alert. While most studies find officers’ race has little or no association with their likelihood of using force (McCluskey and Terrill 2005; Worrall et al. 2021) or receiving an EI alert (Gullion and Ingram 2024), perhaps Hispanic officers take EI interventions more seriously due to their desire to promote to positions held less often by minorities and thus, modify their behavior more so than their counterparts. Additionally, the south, west, and other divisions decreased an officer’s likelihood of a subsequent EI alert, while the south division lengthened their time to a subsequent EI alert. This suggests differences in how supervisors handle EI alerts and interventions across divisions, or how officers embrace or ignore EI interventions with supervisors. Relatedly, Gullion and Ingram (2024) found differences in supervisors’ adherence to EI policy requirements across divisions, including meeting or speaking with officers and other processes. The nature of the officer–supervisor relationship may also play a role in how supervisors handle or officers receive EI interventions. Also, the time to an officer’s subsequent EI alert was reduced if their initial EI alert and intervention were handled by a more tenured supervisor. Perhaps more tenured supervisors feel their ongoing supervision with officers is sufficient, and thus, conducting an EI intervention is unnecessary. Or perhaps given the increased emphasis on police reform and accountability, younger or less tenured supervisors more easily embrace these ideas or have more buy-in for police accountability tools like EI systems or programs, and thus, take the execution of EI alerts and interventions more seriously. If so, this may be perceived by those officers receiving EI interventions as more important to be concerned with modifying their behavior. Gullion and Ingram (2024) also found disconnects between policy and practice in how supervisors were reviewing and executing EI alerts and interventions, including that for each additional year of supervisor tenure, the odds of meeting or speaking with officers or asking officers about personal or job-related stressors significantly decreased.

Third, chargeable vehicle accidents, unconfirmed sick leave, as well as the combination performance indicator decreased an officer’s likelihood of receiving a subsequent EI alert and lengthened their time to a subsequent EI alert. Chargeable vehicle accidents and unconfirmed sick leave may by their nature result in formal or informal disciplinary action, thus, reducing the risk of receiving another EI alert for those performance indicators. Gullion and Ingram (2024) proposed that EI systems may be effective at modifying officer behaviors more willingly changed through supervisors meeting or speaking with officers, which may be true for these indicators. Also, officers who received their initial EI alert and intervention between 2016 and 2020 decreased their likelihood of a subsequent EI alert and had a longer time to a subsequent EI alert. While receiving initial EI alerts later in the study naturally provides less time to receive a subsequent EI alert, perhaps supervisors improved their review and execution of EI alerts and interventions over time. Given this agency’s EI process changes in mid-2017 which reduced summarizing incidents to a recommendation, this may have allowed supervisors to focus more on trends or patterns of behavior (i.e. the bigger picture). Moreover, EI process changes in mid-2017 also required supervisors to ask officers about stressors, which may have also improved officers’ mental health or feelings of supervisor or agency support, modifying their behavior.

Results from this study inform several key policy implications. First, findings reveal the opportunity supervisors and agencies have to provide tailored and appropriate EI interventions with officers, both early and throughout an officer’s career. This may consist of supervisors meeting or speaking with officers about their behaviors and the incidents that triggered the EI alert, reviewing body-worn camera footage related to these incidents, and asking officers about any personal or job-related stressors to address mental health and wellness. Additionally, supervisors could enhance officers’ communication skills and de-escalation techniques and ensure adherence to the tenets of procedural justice through supervision, mentoring, and retraining, especially within the first 2 years of an officers’ initial EI alert and intervention. The opportunity to improve an officer’s performance or their mental health and wellness may exist even when a supervisor determines no formal action is needed, but a pattern or trend can still be addressed (e.g. improving de-escalation techniques for use of force incidents found to be within policy and justified).

Second, given findings may be associated with how supervisors handle the review and execution of EI alerts and interventions, it is critical to hold supervisors accountable to an agency’s EI policy and process requirements through regular oversight by managers. This should consist of reviewing supervisors’ documentation of the EI alert and intervention such as meeting or speaking with officers, discussing incidents or behaviors, and any outcome, whether formal, informal, or no action taken. Furthermore, ensuring EI interventions are handled swiftly and appropriately, including post-intervention monitoring, is vital to preventing repeated misconduct (Walker and Archbold 2013; Gullion and King 2020). Managerial oversight of EI alerts and interventions is also key to understanding supervisors’ decision-making processes and outcomes (Gullion and Ingram 2024). Furthermore, agencies can have those supervisors handling EI alerts and interventions appropriately provide mentoring and training to other supervisors.

Third, findings regarding demographic, occupational, and EI case factors demonstrate that agencies must have a comprehensive and quality EI program. This includes having appropriately defined and regularly assessed EI thresholds to properly identify at-risk officers. Agencies must also have a policy that outlines the EI system’s purpose and processes, including the roles and responsibilities of supervisors and managers, and provide agency-wide messaging and training for EI alerts and interventions. It is equally critical to understand the EI system’s limitations and reasons for the performance indicators and thresholds (Alpert and Walker 2000).

This study has provided valuable insights regarding the likelihood and timing of repeated EI alerts, but it is not without its limitations. Given this study examined data from one police agency, this limits the generalizability. Also, due to the availability of performance indicator data (e.g. uses of force, complaints, chargeable vehicle accidents, etc.), other individual, situational, organizational, and community characteristics affecting the likelihood and timing of EI alerts may not be accounted for. Such factors may include officer and supervisor workgroups, education, prior military experience, supervisor styles and the nature of supervisor–officer relationships, and details of the incidents (citizen demographics, area, outcomes including any formal or informal disciplinary action, etc.), among other factors. Future research should examine additional EI data from other police agencies in other regions and of varying agency sizes and supporting materials of the EI process.

Additionally, this study only examined officers’ initial and subsequent EI alerts that occurred during this study period. Many officers likely had at least one if not multiple EI alerts and interventions prior to the beginning of this study. While an officer’s total number of EI alerts was not considered, officers received up to twelve EI alerts during this study. Future research analyzing officers’ additional EI alerts may offer insight into factors affecting the likelihood and timing of future recurrence of EI alerts, and whether EI interventions are modifying at-risk behavior. This study also only examined EI alerts triggered for sworn officers, to focus on officers who worked in patrol or the field or had regular community interactions. Future research could consider supervisors, managers, and civilian employees, to determine whether and how EI alerts and interventions modified their behavior.

A final limitation of this study is whether the information provided in the supervisor response memos, including the policy and process requirements, accurately represents the experiences of those involved. This includes what officers are aware of and find important regarding EI alerts and interventions, and reasons behind supervisors’ decisions in handling EI alerts and interventions with officers. Supervisor insights were noted in the mixed-methods EI study by Gullion and Ingram (2024), and future research should continue using qualitative methods to examine EI alerts and interventions to gain insight regarding why officers engage in repeated behaviors, the impact of receiving EI alerts and interventions from supervisors, and whether or how it modifies at-risk behavior.

CONCLUSION

The overall goal for police agencies is to address police reform through proper supervision, guidance, and accountability within their organization. This study is a step toward informing agencies how to capitalize on EI systems, a widely recommended supervisory oversight and accountability tool, including how supervisors handle EI alerts and interventions to address at-risk officer behavior. Future research should examine the likelihood of and timing between EI alerts and interventions, repeated misconduct, and other officer activities as this is an unexplored area worthy of examination. The timing component is necessary for the overall picture of the predictors impacting at-risk officer behavior and potential repeated misconduct.

Funding

None declared.

REFERENCES

Footnotes