Why innovations in policing don’t work or don’t translate: An implementation science survey of US police leaders

Abstract

This study examined organizational and community-level barriers and facilitators that influence innovation in policing using an implementation science framework. A body of 31 survey items operationalized for implementation in police settings were mapped onto the constructs of the Consolidated Framework for Implementation Research. The resulting instrument was administered to 72 police leaders and researchers. Respondents agreed implementation bore heavily on an innovation’s prospects for success independent of its effectiveness. There was agreement that innovations in policing were often more complex than the actions they replaced, not integrated into officers’ performance evaluations, and were likely to be mistrusted if they originated from external sources. Officers were unlikely to understand an innovation’s comparative value, and likely to seek their own preferred outcomes regardless. Responses illustrate how effective policing practices could nonetheless be deemed infeasible, providing insights into why many of the innovations that seek to improve policing are met with limited success.

INTRODUCTION

Policing has strong incentives for evidence-based innovation and reform, but these efforts often falter or spread more slowly than they should (Todak & Huey, 2021), and are frequently met with resistance by officers (Kuen et al., 2023; Lum et al., 2012). We do not have a clear understanding of why this is the case, or how to best address these concerns, though recent scholarship suggests that police executives are influential but understudied agents in police reform, given their responsibility for leading organizational change (Ingram, et al., 2022; Adams, et al., 2024). In an effort to improve our understanding of barriers and facilitators of police reform, this study presents data collected using concepts from the flourishing field of implementation science (IS), or “the scientific study of methods to promote the systematic uptake of research findings and other evidence-based practices into routine practice” (Eccles & Mittman, 2006, p. 1). With better knowledge of the barriers and facilitators of organizational change in policing, researchers and practitioners can design and implement innovations in ways that account for these variables, and rigorously study which methods of implementation can help a program, shown to be effective in one setting, facilitate success in others.

Successful innovation in policing can both improve the communities police serve and reduce the harms that result from its work. Officers routinely encounter people in vulnerable physical and mental states, including mental illness, autism, and addiction, and could therefore benefit from a range of partnerships with clinicians and social service workers who possess specialized expertise in handling these encounters. Such collaboration might reduce subsequent use of force and the number of arrests police make, both of which imperil the life course of these vulnerable populations and are the source of strained community relationships. There is evidence, for example, that police can reduce excessive and unnecessary force through changes in policies and tactics (Lopez, 2017), a prospect which has emerging support from randomized trials showing reductions in the use force and subsequently fewer injuries among both members of the public and police (Engel et al., 2020, 2022).

Beyond reducing these undesirable outcomes, policing’s inevitable contact with vulnerable populations could be leveraged to provide critical linkages to preventative care and supports proven more effective at reducing repeat incidents compared to simple arrest and incarceration (Satcher et al., 2024). For example, rather than arresting people who use drugs at the scene of an overdose, which happens with considerable frequency in some jurisdictions (Ray et al., 2022), police could link them to forms of treatment that stand the best chance of reducing the criminal behaviors arising from problematic substance use, such as subsistence crimes and domestic disputes (del Pozo et al., 2023; Reichert et al., 2023). Many of these innovations call for conjoining policing with public health initiatives by pursuing traditional police mandates in ways that either reduce harmful or unproductive collateral consequences, or employ more effective strategies (Goulka et al., 2021), such as police crisis intervention training (Usher et al., 2019) or autism awareness (“Powers and duties of the Board and the Department,” 2024). The thinking is that such approaches, if they are effective, are bound to move beyond punitive responses toward a more habilitative lens, applying person-centered, preventative, and least restrictive practices (del Pozo, Goulka, et al., 2021; van Dijk et al., 2019). In all of these cases, however, the path forward requires implementing new interventions amidst the complex systems of policing.

In addition to this reform-oriented approach, the profession has called for the adoption of evidence-based practices as an unalloyed good in its own right, an endeavor that began over 25 years ago in response to a similar call in medicine (Sherman, 1998). Changes in practice and the use of emerging technologies such as artificial intelligence, drones, and digital forensics might allow police to perform traditional roles more effectively, such as preventing violent crime, solving investigative cases, and safely apprehending wanted suspects. Another frequent goal is increasing community support and perceptions of police legitimacy through enhanced transparency, accountability, collaboration, and performance in an era where police leaders face considerable staffing challenges and elevated levels of scrutiny (del Pozo, Rouhani, et al., 2024). The result has been a flurry of endeavors to bring novel practices and technologies to policing and to gauge the extent they are supported by evidence. For example, the National Institute of Justice has created the Law Enforcement Advancing Data and Science program to support collaboration between police practitioners and researchers, and the American Society for Evidence-Based Policing continues to grow in size and ambition. These advances, accompanied by funding for research and practice from both government agencies (e.g., the Bureau of Justice Assistance) and private philanthropies (e.g., Arnold Ventures and the Robert Wood Johnson Foundation), suggests fertile ground for the evolution of policing.

Actual progress has been slow in many cases, however. Police services in the United States are delivered by over 18,000 agencies, each with its own set of barriers to change, ranging from local political dynamics, to ingrained habits and cultures, to the concerns of labor unions and other interest groups (Skogan, 2014). Some researchers question the positive impact that gradual but persistent improvements in service delivery - without sweeping changes to our social structures – can have on policing and broader justice outcomes in general (Stevenson, 2023). What has been missing from such critiques and assessments, however, is data gathered through the lens of a nascent field of research that studies organizational change itself: implementation science (IS) (del Pozo, Belenko, Taxman, et al., 2024). IS was developed to overcome a similar resistance to evidence-based medicine among physicians (Pope, 2003), where even the most promising findings can take 17 years to find their way into practice (Morris et al., 2011). It endeavors to systematically study the organizational and community-level factors that influence an innovation’s feasibility, acceptability, adoption, penetration, cost, and sustainability (Proctor et al., 2011; Proctor et al., 2023), with an emphasis on identifying the strategies and techniques that maximize the potential for success across these dimensions. To complement the ongoing IS research funded by the National Institutes of Health (NIH/NIDA, 2024; NIH/ODP, 2024), the National Institute of Justice has begun soliciting implementation science demonstration projects and funding programs that require a sustained collaborative partnership between researchers and the practitioners responsible for translating science to practice (NIJ, 2024). These initial steps offer the potential for IS to significantly improve police practice.

Building on a legacy of implementation research

Research relevant to policing has circled IS in a way that leaves it poised to embrace this burgeoning field of study. From the earliest work of pioneers such as Pressman and Wildavsky (1984), scholars have noted how challenging innovation can be in the public sector, examined factors that determine whether policies and programs are successfully implemented, and offered systematic approaches to doing so. For example, Sabatier (1986) presents “top down” and “bottom up” approaches to organizational change, and Sandfort and Moulton (2014) liken overall approaches to innovation to different styles of gardening, using case studies to determine key points in the process of implementation that determine success.

In considering policing, Skogan (2014) notes that a lack of alignment between police leaders, officers’ labor unions, and political actors can impede change, and Schaefer Morabito (2008) finds that the composition of its local government influences a police department’s adoption of community policing strategies. Taking a step back, researchers have examined the factors that determine whether police leaders are receptive to evidence-based practices in the first place, a precondition to successful implementation, and find that rigorous research designs can seem too burdensome to police leaders, who favor less intrusive evaluations of their work (Lum et al., 2012; Telep & Lum, 2014). In examining how internal dynamics influence successful organizational change, Schafer et al. (2021) posit that police supervisors’ messaging plays a significant role in determining whether their subordinates adhere to a new initiative with fidelity (a core concern of implementation science), a finding that accords with other studies that assert police sergeants hold significant sway over the behavioral choices of their officers when a key policy provides them with situational discretion (del Pozo, Sightes, et al., 2021; Marotta et al., 2023). The other factors that determine whether an innovation is implemented with fidelity can be traced back to the storied scholarly lineage of street-level bureaucracy (Lipsky, 1980).

Building on this tradition, implementation science acknowledges these dynamics and proposes to enhance the rigor of their study by carefully operationalizing a wide range of implementation determinants, and providing researchers with the tools to both experiment with different approaches and measure implementation-specific outcomes using a widely-accepted taxonomy of acceptability, adoption, appropriateness, feasibility, fidelity, implementation cost, penetration, and sustainability (Proctor et al., 2011). To succeed, researchers and practitioners need to understand how the theories, models and frameworks of IS can be operationalized and applied in policing, which presently constitutes a major gap in our knowledge (del Pozo, Belenko, Pivovarova, et al., 2024).

The present study

This cross-sectional survey study aims to fill this gap with novel empirical data that police researchers and practitioners can use to bring IS to policing in a rigorous way, using one of its most prominent and widely-accepted frameworks: the Consolidated Framework for Implementation Research (CFIR, 2023). Drawing on the knowledge and experience of police executives from many of the United States’ large municipal police departments, it assesses the extent to which they believe the success or failure of innovations and reforms hinges on careful implementation, and which determinants of implementation pose the most significant challenges. In doing so, it intends to provide the field a scientific approach to testing and comparing implementation strategies and techniques to promote successful innovation and reform throughout policing.

METHODS

This study was conducted utilizing the STROBE checklist for cross-sectional studies as a guide (Von Elm et al., 2007). Determined by Rhode Island Hospital’s Institutional Review Board not to exceed minimal risk, it was designated exempt, with a waiver of written consent (IRB# 2047143-2). All survey respondents received an Explanation of Research prior to data collection.

Study population

The research team surveyed 72 law enforcement employees using a paper instrument. The respondents included police executives, chiefs of police, and a small number of experienced police researchers. Participants were recruited at the annual meetings of the Police Executive Research Forum, the National Institute of Justice’s Law Enforcement Advancing Data and Science Scholars and Academics (NIJ, 2024), and two sessions of the University of Chicago’s Policing Leadership Academy, which educates a national cadre of police executives in evidence-based practices (Crime Lab, 2024). Our inclusion criteria sought respondents who regularly designed, implemented, and evaluated organizational change. Nearly all respondents completed the instrument in person; a handful completed the survey and returned it via mail. We were unable to compute a response rate: while nearly all eligible participants completed the survey in some settings, in others we cannot be sure how many had a feasible opportunity to do so but declined.

Survey development and measures

To create a list of the most relevant determinants of implementation in policing, our survey items were derived from the professional experience of about eight active and retired police executives who were part of the principal investigator’s professional network arising from his service in two municipal police agencies as a uniformed executive. All had suitable backgrounds in upper-level management and directing changes to policies and programs. Responding to prompts from the PI, they were asked which factors, in their experience, held the most influence over successful implementation. The resulting 31 items comprising the survey were organized using the taxonomy of five the domains (Table 1) embodied by the Consolidated Framework for Implementation Research (CFIR) (CFIR, 2023; Damschroder et al., 2022). The CFIR was selected for its intentionally broad array of implementation determinants, 67 in all (Economidis et al., 2023), making it suitable for research in several fields (Kirk et al., 2016).

Table 1.

Domains of the Consolidated Framework for Implementation Research, adapted from CFIR (2023) in (del Pozo, Belenko, Pivovarova, et al., 2024)

Domain	Description	Examples
Innovation	The “thing” being implemented	A new strategy, service, policy, or technology, e.g., facial recognition technology, a de-escalation curriculum, CompStat, drones, etc. Practitioners must ensure an adequate “fit” of the domain to their settings.
Outer setting	The larger settings in which the inner setting exists	There may be multiple levels and outer settings, e.g., local communities and their organizations, city and state governments, etc. Outer settings start with the settings around and above where the innovation takes place
Inner setting	The setting in which the innovation is implemented	Where the innovation will be operationalized, e.g., at the level of the police department, bureau or division, precinct/district, or specialty unit(s).
Individuals	The roles and characteristics of individuals involved with implementing, delivering, and/or receiving the innovation	Headquarters staff, police policy & planning units, precinct/district commanders, supervisors, individual detectives and police officers
Implementation process	Activities and strategies used to implement the innovation	The full range of strategies and approaches used to implement and sustain an innovation, e.g., training, policies, audits, champions, incentives, evaluations, techniques for cultural and behavioral change, etc.

To assess which specific CFIR determinant each operationalized item best corresponded to, three researchers (BdP, ET, and Aw) independently mapped the survey items onto the body of CFIR constructs and discussed discrepancies until consensus was reached. Table 2 displays how each item was determined to map onto the CFIR’s schema and presents its final phrasing and anchoring language. All responses were collected using a five-point Likert scale with anchoring language of either strongly agree and strongly disagree, or never and all the time.

Table 2.

Implementation science survey items mapped onto CFIR domains

Innovation domain		Anchors
Innovation Source	1. These days, when my police agency adopts a new practice, the rank-and-file trust the external actors who helped create it	Strongly disagree/ Strongly agree
Innovation Evidence Base	2. Police agencies don’t adopt new strategies and practices unless there is convincing evidence that they work	Strongly disagree/ Strongly agree
Innovation Relative Advantage	3. When a police department adopts a new practice, the rank and file can be confident is the most advantageous option.	Strongly disagree/ Strongly agree
Innovation Adaptability	4. When a new program or initiative becomes popular, police agencies will adapt it to their own local needs and conditions	Strongly agree / Strongly disagree
	5. When an innovation is brought in from another agency, leaders in the field have flexibility to adapt it to local conditions	6. Strongly disagree/ Strongly agree
	6. One of the advantages of the US police model is that it is uniform enough for innovations to become standardized	Strongly disagree/ Strongly agree
Innovation Trialability	7. The police profession is good at implementing pilots with a plan to expand and sustain successful ones department-wide	Strongly disagree/ Strongly agree
Innovation Complexity	8. When alternatives to enforcement are introduced, they are more difficult to execute than the original enforcement practice	Never/ All the time
Outer Setting domain
Local Attitudes	9. In my experience, police agencies spend adequate time gaining the support of community leaders for new technologies	Strongly disagree/ Strongly agree
Local conditions	10. Communities understand how to support new police practices to increase their likelihood of success	Never/All the time
Policies & Laws	11. When police agencies want to stop a practice, they believe a suitable rule or regulation is enough to do so successfully	Never/All the time
	12. New police practices are accompanied by the changes to practice, procedure and law that help make them successful	Never/All the time
External Pressure	13. Community leaders and elected officials tell police to adopt a program without knowing how it can be implemented	Never/All the time
Societal & Market Pressure	14. External pressures affect the new practices police adopt and the timelines for their implementation	Never/All the time
Inner Setting domain
Info Technology Infrastructure	15. When we start a program or initiative, it is accompanied by the necessary changes in IT to ensure high performance	Never/All the time
Communications	16. When we innovate there is a way to communicate feedback to a person who will listen and has the power to make changes	Strongly disagree/ Strongly agree
Culture	17. We have buy-in from the rank and file for innovations and new programs because they clearly understand their values	Strongly disagree/ Strongly agree
	18. It is police culture that determines if a given practice will cease more so than the decisions made by police executives	Strongly disagree/ Strongly agree
Incentive Systems	19. When an agency provides an alternative to arrest, police leaders explain how it will be used in performance evaluations	Never/All the time
Mission Alignment	20. If police officers strongly prefer a certain outcome, they will find ways to get that result regardless of changes to procedures	Strongly disagree/ Strongly agree
Individual Characteristics domain
High-level Leaders	21. The police leaders I know think successful implementation is more of an art than a science	Strongly disagree/ Strongly agree
Opinion Leaders	22. Police agencies identify and encourage the champions of new practices to promote the practice's legitimacy and success	Never/All the time
Innovation Deliverers	23. When my agency implements a new practice, people know who in their chain of command is responsible for its success	Never/All the time
Process domain
Teaming	24. When police innovate, it is typical for all levels of command, local government, and the community to act as a team	Strongly disagree/ Strongly agree
Assessing Needs of Innovation Deliverers	25. I have seen a police program or initiative fail because it wasn’t implemented the way I thought was necessary	Never/All the time
	26. The first time rank and file police officers hear about a new practice or innovation is when they get word it is in effect	Never/All the time
Planning	27. Police departments publicly announce programs and initiatives before planning their effective implementation	Never/All the time
	28. When police in my agency have a question about a new initiative or program, they knew who they can turn to for answers	Never/All the time
Tailoring Strategies	29. I have seen a detailed plan to implement a new evidence-based police practice, expand it, and sustain it over time	Never/All the time
Engaging	30. In my experience, rank and file police officers are briefed on their role in making a new program successful	Strongly disagree/ Strongly agree
Reflecting & Evaluating	31. Poor implementation is the reason why evidence-based practices fail in a given police agency	Never/Every time

Data were collected from May 2023 through March 2024, concluding with the administration of the survey to the second cohort of the University of Chicago’s academy. The data were then compiled and cleaned. Of a possible 2,232 responses, 46 were missing (2.0%; mean per item=1.5, max=4, SD=1.3), and there were no patterns visually detected in these few omissions. As an exploratory study of a novel research subject in policing, analyses were limited to descriptive statistics and computing the mean and standard deviation of the responses for each item, which did not require pair or listwise deletions for missing data.

RESULTS

The study’s 72 respondents were police executives and a handful of colleagues (see below) who closely collaborate with them on innovation in major police departments including those of New York, Chicago, Philadelphia, Boston, San Francisco, and Los Angeles, as well as agencies serving other municipalities, with 38.9% of respondents serving in an agency of at least 1,000 sworn officers. They were sampled across all Census Bureau regions, with 37.7% from the South, 24.6% from the West, 20.3% from the Midwest, and 17.4% from the Northeast (see Table 2). Most respondents served in urban areas (87.0%) and held a rank of Captain or above (80.6%). Of the remainder, nine respondents were a sergeant or lieutenant, one was an officer or detective, and four were non-sworn police employees. Among the 72 police executives surveyed, educational attainment was heavily skewed toward advanced degrees: 2.8% held associate degrees, 32.4% held bachelor’s degrees, 49.3% held master’s degrees, and 8.5% held a PhD, which included four policing researchers, and two uniformed police officials. The researchers and rank-and-file officers included in the sample were invited to participate due to their considerable experience collaborating with police executives in designing, implementing and evaluating innovations and reforms, providing insights into the process on par with their uniformed executive colleagues.

The majority were seasoned professionals, with 51.4% reporting a tenure of 16 to 25 years, and 38.9% reporting over 25 years of service. Complete demographic data for respondents can be found in Table 3.

Table 3.

Demographics of implementation science survey respondents, n=72

Census Bureau region	%
South	37.7
Midwest	20.3
West	24.6
Northeast	17.4
Location
Rural	13.0
Urban	87.0
Rank
Officer or Detective	1.4
Sergeant or Lieutenant	12.5
Captain or above	80.6
Non-sworn employee/researcher	5.6
Highest level of education
High school	7.0
Associate’s degree	2.8
Bachelor’s degree	32.4
Master’s degree	49.3
J.D.	0
Ph.D.	8.5
Respondent tenure
8-15 years (mid-career)	9.7
16-25 years (senior)	51.4
More than 25 years	38.9
Sworn agency head count
0-99	16.7
100-499	29.2
500-999	15.3
1000+	38.9

Respondents overwhelmingly believed that the process of implementation bore heavily on an innovation’s prospects for success and that this relationship was independent of the effectiveness of the innovation itself. We describe the most notable responses by domain below, with results for all 31 items presented in Table 4.

Table 4.

Response analysis by experimental item, n=72

CFIR Domain	Survey Item	Response %			Mean	SD
		1-2 (Generally disagree)	3 (Neutral)	4-5 (Generally agree)
Innovation	1. Cops trust innovation source	73.6	11.1	15.3	2.1	1.1
	2. Agencies adopt policies with a sufficient evidence base	47.2	22.2	30.6	2.8	1.2
	3. Cops trust an innovation is the most advantageous option	62.9	24.3	12.7	2.4	0.9
	4. Agencies will adapt innovation to own local needs and conditions	55.6	25.0	19.4	3.4	1.0
	5. Leaders in the field have flexibility to adapt innovation to local conditions	28.6	28.6	42.9	3.2	1.0
	6. US police model is uniform enough for innovations to become standardized	50.0	35.7	14.3	2.4	1.1
	7. Agencies are good at innovation piloting/trialability	61.4	24.3	14.3	2.5	0.9
	8. Innovation is more complex than original practice	14.7	23.5	61.8	3.6	0.9
Outer Setting	9. Agencies spend adequate time gaining community leader support for tech innovation	73.6	11.1	15.3	2.6	1.0
	10. Local conditions support new practices	47.2	22.2	30.6	2.2	0.8
	11. Agencies believe they can stop practices with policies & laws alone	62.9	24.3	12.7	3.4	1.0
	12. New practices accompanied by necessary changes to policies & laws	55.6	25.0	19.4	3.2	0.9
	13. Societal pressures push police to adopt programs	28.6	28.6	42.9	3.9	0.9
	14. External pressures affect the new practices and their timelines for implementation	50.0	35.7	14.3	3.9	0.9
Inner Setting	15. Innovation is accompanied by the necessary structural changes in IT	55.7	24.3	20.0	2.6	1.0
	16. Innovation comes with adequate communication	37.1	28.6	34.3	3.0	1.0
	17. Buy-in from cops for innovations because they clearly understand their value	58.6	18.6	22.9	2.6	1.0
	18. Police culture determines if a practice will cease	8.8	26.5	64.7	3.7	0.9
	19. Leaders explain how alternatives to arrest will be used in performance evaluations	81.4	11.4	7.1	1.9	0.9
	20. Officers will find ways to get desired result regardless of changes to procedures	13.9	43.7	42.3	3.6	1.0
Individual Characteristics	21. Police leaders think successful implementation is more of an art than a science	14.1	43.7	42.3	3.4	0.9
	22. Agencies encourage champions of new practices to promote its legitimacy and success	30.0	40.0	30.0	3.0	0.9
	23. People in agency know who is responsible for an innovation’s success	35.2	21.1	43.7	3.1	1.1
Process	24. When police innovate, teamwork is typical	65.7	22.9	11.4	2.4	1.0
	25. A practice failed because wasn’t implemented the way I thought was necessary	8.5	21.1	70.4	3.8	0.9
	26. Assessing needs of innovation deliverers - lack of communication until commencement	21.4	27.1	51.4	3.4	1.1
	27. Police publicly announce programs before planning implementation	29.2	33.3	37.5	3.1	1.1
	28. Officers know who to turn to for questions of innovation	40.3	18.1	41.7	3.1	1.1
	29. I have seen a detailed plan w/ tailored strategies for implementation and sustainment	29.0	26.1	44.9	3.2	1.3
	30. Officers are briefed on role in making new program successful	54.2	12.5	33.3	2.8	1.2
	31. Poor implementation is why evidence-based practices fail in a given agency	2.9	33.3	63.9	3.7	0.7
Responses presented as a visual analog scale with anchoring language

Innovation.

The police leaders in this study assessed statements regarding an innovation’s source, evidence base, relative advantage, adaptability, trialability, and complexity. When a new practice was introduced to their agency by an external source, 73.6% of respondents disagreed that “the rank-and-file trust the external actors who helped create it.” A majority (61.4%) disagreed that policing was “good at implementing pilots with a plan to expand and sustain successful ones department-wide,” with only 14.3% agreeing, and the rest giving a neutral response. Regarding the increased complexity that can come with change, 61.8% agreed that “when alternatives to enforcement are introduced, they are more difficult to execute than the original enforcement practice.” While only 14.3% of respondents agreed that policing has a uniformity that allows for the standardization of an effective approach, only 19.4% agreed that agencies will take the time to tailor a popular innovation to their local needs and conditions.

Outer Setting.

Respondents evaluated statements regarding the external environment that the inner setting exists and, in this case, the community and governmental bodies that the police departments serve. Most respondents (73.6%) disagreed that agencies “spend adequate time gaining the support of community leaders for new technologies,” that “new police practices are accompanied by the changes to practice, procedure and law to make them successful” (55.6%), while most respondents (62.9%) disagreed that police agencies believed changes to policies or laws alone were enough to stop an undesirable and ineffective practice.

Inner Setting.

Respondents expressed their beliefs about the implementation environment within their police department. The most universal sentiment among respondents (81.4%) was that when an agency provides an alternative to arrest, police leaders rarely explain how it will be used in performance evaluations. In contrast, few respondents (13.9%) disagreed with the problematic idea that “if police officers strongly prefer a certain outcome, they will find ways to get it regardless of changes to procedures,” which was congruent with the widely-held belief that “it is police culture that determines if a given practice will cease more so than decisions made by police executives” (64.7% agreement). Compounding the potential obstacles this suggests, a majority (58.6%) of respondents disagreed that “we have buy-in from the rank and file for innovations and new programs because they clearly understand their value,” and that new initiatives are “accompanied by the necessary changes in IT to ensure high performance” (55.7%).

Characteristics of Individuals.

Respondents were presented with items regarding high-level police leaders, leaders with informal influence, and those who might deliver an innovation. There was little disagreement (14.1%) that police leaders thought “successful implementation is more of an art than a science,” and they were evenly split (30% agreeing or disagreeing; the remainder giving a neutral response) that “police agencies identify and encourage the champions of new practices to promote their legitimacy and success.” Fewer than half agreed that people in their chain of command were ultimately responsible for ensuring that success (43.7%).

Process.

Finally, participants rated statements regarding the strategies utilized to implement the innovation such as teaming, assessing needs of the innovation deliverers, planning, tailoring strategies, engaging, and reflecting and evaluating. Few respondents (11.4%) felt that “when police innovate, it is typical for all levels of command, local government, and the community to act as a team,” and most (54.2%) disagreed that “rank and file police officers are briefed on their role in making a new program successful.” Regarding why innovations ultimately fail, 70.4% of the respondents believed they were not implemented in a way they thought was necessary. Further, when “evidence-based practices fail in a given police agency,” 63.9% of respondents felt poor implementation was likely to blame, and only 2.9% felt that was unlikely to be the case.

Exploratory bivariate associations

The size of our sample, the exploratory nature of this study, and the novelty of its survey items indicate caution in subjecting our dataset to complex analyses. This was our rationale for emphasizing the study’s descriptive results, which can serve as the basis for more extensive, rigorous research. In order to provide further potential guidance for such studies, we also conducted pairwise analysis of the items in our survey. The full results are presented in Table 5. Among the results that were significant at P<.05, some correlations were notable. The strongest was a relationship between the belief that police agencies identify internal champions to promote an innovation’s legitimacy, and that an agency offers the means for officers to provide feedback to a person who will listen and respond to the concerns (r = 0.64). Relatedly, there was an association between the belief that agencies employed champions for an innovation and that the an agency’s rank and file supported innovations because they clearly understood their value (r = 0.52). There was also a significant association between the belief that a communities understand how to support new police practices to increase their chances of success, and the belief that rank-and-file officers trust the external actors who helped create a practice (r = .57). In a similar vein, the belief that the community, local government, and police act as a team when agencies innovate was also associated with rank-and-file trust in the external actors involved in innovation (r=.55). Finally, the belief that there was an effective way to provide feedback about innovation to a person empowered to act on it was correlated with a belief that officers can be confident a given innovation is the most advantageous option (.57). These preliminary analyses suggest that steps to ensure careful implementation in a police setting’s outer domain associate with implementation outcome variables inside the agency, and key actors within an agency that act as champions and conduits for feedback can be leveraged to improve the odds of successful implementation.

Table 5.

Correlations across CFIR constructs (numbers correspond to items listed in Tables 2 and 4) Note: * p < .05

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30
2	0.01	—
3	0.01	.29*	—
4	−0.03	.43*	0.15	—
5	0.04	.38*	.48*	0.23	—
6	0.09	.45*	.30*	.27*	.27*	—
7	0.21	0.2	−0.04	0.14	0.01	0.2	—
8	−0.11	.43*	0.19	.42*	0.24	0.19	0.01	—
9	0.24	−0.05	−0.1	−0.01	0.05	−0.01	0.03	−0.2	—
10	−0.08	−0.07	0.15	−.28*	−0.19	0.04	0.07	0.06	−0.01	—
11	0.01	.29*	.30*	.38*	.27*	0.16	−0.02	.27*	−0.06	−.27*	—
12	−0.02	.52*	0.15	.54*	0.16	.25*	.30*	.28*	0.01	−0.08	.37*	—
13	0.03	0.18	−0.12	−0.1	0.2	−0.08	−0.05	−0.06	0.13	−0.13	−0.22	−0.07	—
14	0.01	0.07	−0.09	−0.17	0.08	0.1	−0.06	0.08	0.07	0.22	−.27*	0	0.14	—
15	.45*	0.01	0.06	−0.05	0.08	0.05	0.19	−0.17	.48*	0.18	0	0.01	−0.04	−0.07	—
16	0.15	0.2	31*	0.2	.43*	.27*	0.1	0.22	0.01	−.31*	.32*	0.12	−0.13	−0.01	0.11	—
17	0.15	−0.22	−0.23	−0.23	0	−0.09	0.02	−.27*	0.16	0.06	−.41*	−.27*	019	0.08	.26*	−0.12	—
18	0.06	.46*	0.19	.59*	.28*	.46*	.29*	.39*	−0.08	−0.22	.57*	.50*	−0.04	0.07	−0.06	.35*	−.30*	—
19	0.01	.35*	−0.15	.28*	0.06	0.12	.30*	.34*	0.03	−0.02	0.24	.42*	0.09	0.1	−0.08	0.11	−0.13	.41*	—
20	−0.06	.30*	0.1	.43*	0.18	.33*	0.24	0.23	−0.05	−0.12	0.15	0.24	0.05	−.29*	0.01	0.14	−0.17	.29*	.35*	—
21	.43*	−0.01	−0.01	0.05	0	0.06	−0.07	−0.11	.35*	−0.02	−0.03	−0.07	−0.04	0.02	.30*	.27*	.27*	−0.04	−0.1	−0.12	—
22	0.03	.32*	.36*	.43*	.26*	0.17	0.12	.33*	−0.09	−0.16	.34*	.29*	−0.11	−0.06	0.06	.43*	−0.19	.31*	0.15	.33*	−0.02	—
23	0.03	.39*	.41*	.37*	.45*	.35*	0.01	.49*	−0.1	−0.06	.29*	.35*	0.07	0.1	−0.01	.39*	−0.18	.46*	.38*	.37*	−0.1	.42*	—
24	0.06	−0.1	−0.05	−.38*	−0.08	−0.05	−0.01	0.02	0.08	.26*	−0.03	−0.11	−0.03	0.11	0.2	−0.24	0.17	−0.22	−0.06	−.36*	0.16	−0.14	−0.08	—
25	−0.05	.53*	0.21	.42*	0.12	0.11	0.1	.38*	−0.16	−0.2	.51*	.39*	−0.12	−0.17	−0.06	0.19	−.44*	.42*	.31*	.31*	−0.08	.41*	.42*	−0.15	—
26	0.05	.38*	0.04	.40*	0.16	.32*	0.05	.27*	−0.06	−0.12	.36*	.45*	0.06	−0.07	−0.06	0.2	−.31*	.47*	.30*	0.21	−0.15	0.17	0.19	−0.17	.36*	—
27	−0.01	.64*	.33*	.47*	.32*	.46*	0.01	.31*	0.12	−0.08	.33*	.50*	0.04	−0.02	0.05	.28*	−0.17	.44*	.34*	.52*	0.02	.35*	.57*	−0.19	.46*	.35*	—
28	0.03	0.06	−0.04	−.40*	−0.03	0.11	0.12	−0.13	0	0.16	−.28*	−0.17	0.12	.24*	0.14	−0.16	.29*	−.28*	0.11	−0.09	0.02	−0.23	−0.1	.25*	−0.12	−0.19	−0.1	—
29	−0.1	.49*	.32*	.36*	.33*	.30*	0.05	.34*	−0.1	−0.15	.47*	.31*	0.06	0.01	−0.05	.30*	−0.21	.55*	0.07	0.2	−0.02	0.22	.41*	−0.1	.44*	0.14	.42*	−0.23	—
30	0.18	−.25*	−.27*	−0.21	−0.21	−0.15	−0.11	−0.12	.32*	0.1	−.35*	−0.21	0.02	0.14	0.19	−0.18	.39*	−.25*	−0.04	−.32*	.31*	−.40*	−.25*	.31*	−.39*	−0.22	−0.22	−0.03	−0.2	—
31	0.03	0.09	0	−0.01	0.17	−0.07	.28*	0.17	−0.06	0.08	0.05	0.13	0.15	0	−0.04	0.6	−0.01	.26*	.25*	0.03	−0.14	0.01	0.22	0.11	−0.02	0.08	0.12	−0.2	−0.04	0.03

DISCUSSION

This is, to our knowledge, the first survey of executives and other key police personnel about the determinants of successful implementation in their profession. In presenting descriptive empirical data about these determinants, the results highlight a critical gap in our approach to organizational change in policing, and provide actionable insights into why many innovations and reforms in policing are met with limited success despite a promising evidence base. Our findings highlight the need for research into the role and performance of police executives, given how their attitudes toward organizational change and their approach to implementation can determine the success of promising reforms (Adams et al., 2024; Matusiak et al., 2019; Mourtgos et al., 2024). Given respondents’ views about implementation, we can see how an intervention with strong internal validity could nonetheless fail. The responses suggest it could be deemed infeasible, lack acceptance among officers, not be adopted for use despite changes to training and policy, prove to be unsustainable, or fall short of the other outcomes that IS researchers use to measure success (Proctor et al., 2023).

It is instructive to note, for example, that there was widespread concern among the police leaders in this study that alternatives to enforcement are more complex than the actions they seek to replace. In addition, they may not be supported by the necessary changes to an agency’s IT, not articulated as a clear component of employee performance evaluations, and are likely to be mistrusted by police if they originated from external sources. Further, officers are unlikely to understand their comparative value and will try to find a way to get the outcomes they prefer regardless. This is of significant concern when police encounter vulnerable health populations such as those with mental illness, developmental disabilities, or who engage in problematic substance use. Evidence shows that such individuals and their families avoid requesting support from law police due to a fear of exposure to excessive force or arrest. For many, these negative experiences are driven by a lack of police awareness of signs and symptoms of their condition, safe procedures to de-escalate, and viable alternatives to arrest (Young & Brewer, 2020).

As a concrete example of how our findings can be used by researchers, consider efforts by scholars to understand the impact of arrest diversion programs. If a reform aimed at arrest diversion is more complex than an arrest, is not accounted for in formal evaluations, and officers are not given a compelling justification – IS would predict the innovation, no matter how carefully designed, is at a high risk of failing as a result of how it was implemented. Each of these determinants presents a formidable barrier to successful implementation, and all it may take is a few of them to manifest modestly to produce a compounding effect that derails an innovation. Similarly, police acceptance of body cameras, to include adopting their use and complying with policies that call for their activation, has been shown to influenced by the extent to which officers realize the innovation serves not only a higher strategic goal (such as collectively modifying officer behavior), but is useful to their own personal, officer-level ends as well, thereby fostering a form of pragmatic legitimacy that can be crucial for successful implementation (Young & Ready, 2015). This suggests we lack a thorough understanding of the extent to which an innovation or reform is suboptimal or ineffective: because it inherently lacks causal efficacy, or because barriers to successful implementation have reduced its power either during initial deployments or upon its translation to other settings.

While the respondents of this survey took a markedly dim view of the quality of implementation in their field, this is not a self-assessment unique to policing. IS arose as a field because physicians were highly skeptical of emerging evidence-based practices, their institutions were often resistant to organizational change (Pope, 2003; Sherman, 1998), and many valuable and well-tested improvements to care have languished without ever being successfully implemented at all (Kirchner et al., 2020). Despite efforts to ameliorate these concerns and others, some of its pioneers view the movement toward evidence-based medicine as verging on a state of crisis (Greenhalgh et al., 2014). Very few professions that deliver complex services directly to people in dynamic, resource and time-constrained environments have taken a scientific approach to understanding how they can successfully implement change, policing included.

It is beyond the scope of this article to extensively discuss how the data presented here can be leveraged to increase the practice of IS in policing, but a few observations are warranted. The first is that evaluations of police practices should exercise caution in drawing conclusions about their effectiveness or translatability unless they have probed the influence implementation had on outcomes. Repositories that collect these evaluations, such as the Evidence-Based Policing Matrix (Lum et al., 2022), should note the limits of evaluations that do not do so, and explicitly advise readers that null results may not necessarily indicate an innovation lacks efficacy. Otherwise, policing risks stasis, where the traditional methods of evaluation will “provide little or no clue as to why the intervention worked or did not work when applied in different contexts or circumstances, deployed by different stakeholders, or used for different purposes” (Pawson et al., 2005, p. S1:21).

Efforts to translate a successful practice to a new setting should therefore reflect that its success may hinge on identifying and overcoming barriers to implementation, which include not only a lack of fidelity to the innovation’s design, but the failure to modify it to accommodate local conditions. Using the pilot data provided here as a guide, police executives could pay particular attention to the barriers and facilitators to implementation present in their own agencies. Researchers could design approaches seeking regular, ongoing feedback from targeted adopters to ensure ongoing acceptability and appropriateness, and test the comparative effectiveness of different approaches to implementing a program once it is developed. In doing so, hybrid trials that combine studies of implementation and effectiveness (Landes et al., 2019) may be well-suited for the dynamic operational environments found in policing (del Pozo, Belenko, Pivovarova, et al., 2024).

As IS in policing evolves into a more specialized field, it would be helpful for the profession to adapt and tailor CFIR-based determinants for use in a police-specific framework. There is likely no need for additional theories, models, and frameworks in IS for use in policing, however. Many exist already, and have proven their worth in professions of similar size and complexity as policing. At the same time, a plethora of approaches also suggests IS is an immature science in the process of consolidating its theoretical tools into a small group that offers the greatest rigor and most consistent explanatory power (Hibbert, 2016). If true, there is no need to slow down this process by complicating such a consolidation with additional items.

That said, we may find there are unique aspects of implementation in police settings that are persistent enough to be considered widely applicable determinants. Policing’s constellation of subcultures and its specific mission mean certain aspects of successful organizational change are likely to be pronounced, persistent, and worth focusing on as IS research takes root in policing. Such a body of common barriers and facilitators can serve as a rigorous foundation for IS research in the field, provided that it is moored to something like the CFIR and operationalized by a suitable model or framework. The items and data presented in this exploratory study intend to provide a starting point for this endeavor.

The preliminary data here can help shape a research agenda for implementation science in policing. The most immediate research can confirm the salience of the implementation variables presented here across different police settings, more rigorously analyze the bivariate relationships observed here, and further operationalize the constructs of the CFIR to develop new items. Evaluations can then examine not only the effectiveness of a program or innovation, but collect retrospective data about discreet aspects of implementation and their resulting outcomes. As discussed in del Pozo, Belenko, Pivovarova, et al. (2024), the final frontier of such a research agenda would be the design of experimental implementation trials which would prospectively employ different approaches to the implementation of a given innovation and assess them in terms of the prevailing taxonomy of implementation outcomes (Proctor et al., 2011).

Limitations

This study has limitations. It consists of a convenience sample of 72 respondents who were only asked to assess the determinants presented to them in the survey. This limits the study’s generalizability, and leaves open the possibility it failed to identify potentially influential determinants. The data are descriptive, an approach that lacks nuance and does not shed light on the relationships between the variables we examined, which would have required a larger sample to supply the necessary statistical power. Further, the survey instrument was limited to closed-form items, a well-established survey practice that scholars have nonetheless critiqued for imposing researchers’ expectations on the data and thereby limiting the potential responses (Mourtgos & Adams, 2019). In the case of our study, these items were constructed with extensive input from respondents’ peers, significantly raising the likelihood that they captured salient constructs.

These limitations were accompanied by corresponding strengths, however. The participants served several major US police departments, and were recruited from venues that typically draw police executives amenable to—but more often genuinely committed to—organizational change, and who are of sufficient rank and experience to offer accurate insights about it. For example, the police leadership program at the University of Chicago was designed for executives selected by their agencies for their promise as organizational innovators, and the LEADS program carefully selects participants based on their willingness to study and implement evidence-based practices. While a larger sample might allow for more confident generalization or permit more sophisticated analyses, it is unclear that sampling participants from across a broader range of U.S. police agencies would have improved the quality of the responses, or the insightfulness of the participants. Despite the study’s limitations, our findings from experienced, high-ranking, and highly educated participants can provide a reliable foundation for future IS research and experimentation. Future research, however, should also solicit participation from rank-and-file officers to determine if their attitudes and beliefs about organizational change are congruent with those of the executives who design and implement it.

CONCLUSIONS

By operationalizing selected determinants of the Consolidated Framework for Implementation Research, this cross-sectional survey study reports on respondents’ beliefs about the barriers and facilitators of successful implementation in policing. Police agencies can use it as a guide for planning purposes, and researchers can use it as the preliminary basis for IS studies in a profession that has yet to see a prospective one take place. As the demand for innovation in policing continues without relent, agencies will come under increasing scrutiny from the people and institutions in the communities they serve if they are unable to leverage the growing number of evidence-based practices they are meant to deliver. To prevent this, implementation science can provide police departments with a roadmap to more successful and sustainable organizational change, grounded in an awareness of the determinants that make a new practice tenable independent of how compelling the evidence is that it “works.”

Biographies

Brandon del Pozo, PhD, MPA, MA studies public health, public safety, and justice at Brown University

Rose E. Nevill, PhD researches police approaches to people with autism at the University of Virginia

Javier Cepeda, PhD, MPH is an NIH-funded researcher at the Johns Hopkins Bloomberg School of Public Health who conducts research that aims at improving the health outcomes of police encounters with people who use drugs.

Ian T. Adams, PhD studies policing at University of South Carolina

Alina Whiteside, BS is a research assistant at Brown University Health

Erin Thompson, MPH is a research project coordinator at Brown University Health