The Effects of Body-Worn Cameras on Self-Initiated Police Encounters

Abstract

Beginning in the summer of 2014, a series of controversial deaths in the US, often involving white officers and unarmed minority men, contributed to the growing debate about the need for increased transparency and accountability in policing. Central to the debate has been the call for more body-worn cameras (BWCs), a widely adopted technology that promises numerous benefits through real-time recording of police-public encounters. However, critics of this technology claim that increased surveillance of their behavior may lead to de-policing where officers choose to reduce the quality and/or quantity of their self-initiated police work. This claim has recently received empirical scrutiny in the context of the two main self-initiated activities underlying police work: foot patrols and traffic stops. However, the studies have mainly been outcome-orientated (e.g., stop and frisk or issuance of a citation) rather than activity-focused. The de-policing claim is tested here using data from a lengthy, quasi-experimental evaluation of BWCs in a large US police department. The findings show that the number of foot patrols and traffic stops conducted did not vary based on whether an officer wore a BWC or not. Policy and theoretical implications of these findings are discussed.

Keywords

police body-worn cameras de-policing technology

Introduction

Technological innovations have long been associated with advances in modern policing and public safety (c.f. Lynch, 2015 for a review) and for the better part of the last decade, police departments across the US have considered or adopted body-worn cameras (BWCs). Their use has accelerated dramatically in recent years because of numerous high-profile and controversial deaths of community members during interactions with the police.

One of the first of these high-profile incidents involved the shooting death of Michael Brown in 2014 by a Ferguson, Missouri, police officer.¹ This was followed by the deaths of Eric Garner, Tamir Rice, Walter Scott, Freddie Gray and many others. More recently, the deaths of George Floyd and Breonna Taylor in 2020 made national headlines and sparked a series of large protests across the country, which many observers believe led to the 2021 convictions of one police officer for the George Floyd murder and another for manslaughter in the 2021 shooting of Daunte Wright.

In light of these deaths, public outrage began to surface that centered on police use of force, the disparate treatment of minority populations and the need for stronger oversight of police actions. Many of these same concerns were highlighted, presciently, in the final report of the President’s Task Force on 21st Century Policing (2015), which pointed specifically to the growing sense of distrust and lack of police legitimacy within minority communities. The Task Force provided approximately 60 recommendations that addressed these problems, including the adoption of BWCs.

At the same time, community groups began to demand that police departments deploy BWCs in hopes that by recording interactions between police and the public, there would be greater officer accountability, transparency and thus enhanced police legitimacy (see general discussions by Lum et al., 2019; Maskaly et al., 2017; Nowacki & Willits, 2018; White, 2014). Indeed, previous studies have shown early and widespread public support for the adoption of BWCs in the US (Crow et al., 2017; Sousa et al., 2018; White et al., 2017).

It is within this context that the adoption of BWCs has become a focal issue of police reform nationally. In 2013, the Bureau of Justice Statistics’ Law Enforcement Management and Administrative Statistics (LEMAS) survey estimated that nearly one-third of agencies had outfitted patrol officers with BWCs (Reaves, 2015). Just 3 years later, this number rose to nearly half (Hyland, 2018). As further indication of the growing use and importance of BWCs, many government and professional organizations such as the US Bureau of Justice Assistance, the International Association of Chiefs of Police and the Fraternal Order of Police developed model policies for the adoption and implementation of the technology. Currently, seven states have passed laws mandating that all law enforcement officers who have contact with the public must be equipped with BWCs by 2023 (Fredericks, 2021).

An increasing number of studies show that the technology provides beneficial outcomes, including reductions in complaints against officers (Braga et al., 2018a, 2018b; Goodison & Wilson, 2017; Grossmith et al., 2015; Hedberg et al., 2017; Jennings et al., 2015; Katz et al., 2014) and heightened perceptions of procedural justice among the public (White et al., 2017). However, observers have questioned whether the adoption of BWCs may also produce unintended consequences such as reduced public safety through de-policing, where officers respond to the increased oversight by reducing the number and intensity of their self-initiated, non-dispatched law enforcement actions (Davenport, 2017; MacDonald et al., 2016; Rosenfeld, 2016; Rushin & Edwards, 2016). Such claims have been subjected to some rigorous research, but as Braga et al. (2017, p. 13) note, “… it has become clear that the technology’s potential implications and impacts are far-reaching and not widely understood.”

Community leaders, police and other public officials that have advocated for BWCs have done so without knowing the extent to which the technology might result in de-policing which we define as a decline in the quality and quantity of police officers’ self-initiated work. The decline would result in a weakening of officers’ ties to the neighborhoods they patrol and thus a reduction in the public safety they deliver. Most prior studies have focused on specific outcomes of police work like stop and frisk encounters with the public and the issuance of traffic citations, outcomes that are easily measured and routinely captured in police databases. However, even when combined, the two outcomes are insufficient measures of self-initiated police activities because they miss many other types of valuable services such as crime prevention, intelligence and information gathering, fostering community relations, showing a police presence and the issuance of verbal warnings. This study tests the de-policing hypothesis using more encompassing measures of the two major activities of self-initiated policing, foot patrols and traffic stops.

The study reported here is based on 96 weeks of data collected by the Fairfax County Police Department, VA (FCPD), as it conducted a lengthy evaluation of a pilot BWC-deployment program during 2018–19. The evaluation used a quasi-experimental design to compare the number of times an officer on foot or traffic patrol duty chose to leave their cruisers for reasons other than a dispatched service call, i.e., conducted self-initiated policing. The data were analyzed using an interrupted time-series analysis tool (itsa) well-suited to the study’s design and data collection protocol.

Review of the Literature

The literature on BWCs and their impact on police conduct, transparency, and accountability has expanded dramatically over the past several years. In 2019, Lum and associates (Lum et al., 2019) identified 70 empirical studies that addressed one or more of four topics: the impact of BWCs on officer behavior, officer attitudes about BWCs, the impact of BWCs on citizen behavior, and citizen and community attitudes regarding BWCs. The officer behaviors of interest included measurable outcomes such as citizen complaints, use of force reports and arrests. The researchers also examined officers’ self-initiated, discretionary actions such as vehicle stops and pedestrian stop and frisks while on patrol. The authors concluded that the 70 studies offered little or mixed support to hopes that BWCs would improve officer relations with their communities.

A year later, Lum and her research team (Lum et al., 2020) published a meta-analysis of a more limited set of 30 studies which employed rigorous research designs (either random controlled trials or quasi-experimental designs) that focused on officer or the publics’ behaviors. The majority of these studies focused on use of force (26 studies) and civilian complaints against police officers (22 studies). Of the 30 studies, eight addressed self-initiated police behaviors measured in a variety of ways, four also examined traffic stops and four also measured field interviews or stop and frisks. Lum and associates found no statistically significant results in their meta-analyses of these eight studies. They noted the likely overlap between various measures of self-initiated police behavior and measures of specific patrol outcomes and thus the need for further research on this topic.

Two social psychological perspectives have been used to explain why de-policing should occur when BWCs are present: deterrence and self-awareness (Braga et al., 2018b; Groff et al., 2020). The deterrence perspective maintains that an officer will avoid activities that might risk confrontations with civilians, which if handled improperly, could lead to sanctions against the officer. This perspective is most salient when an officer is equipped with a BWC and their actions are being recorded, a situation wherein an officer’s word need not be taken reflexively over that of the public. If the deterrence effect is operative, camera-wearing officers should engage in fewer self-initiated, risk-creating behaviors than officers who are not.

The self-awareness perspective is hypothesized to generate the same de-policing effect as the deterrence model but does not involve consideration of external sanctions. It is based on the simple notion that when someone believes their actions are being observed, even if only by a camera, they tend to behave according to their understanding of applicable social norms. In this context, the hypothesis rests on the assertion that few officers want to be seen deviating from the occupational standards they have internalized and the departmental policies they have sworn to uphold (Braga et al., 2018b; Ready & Young, 2015). Again, if the self-awareness effect is operative, officers equipped with BWCs should initiate fewer contacts with the public because of the risk that interactions might trigger behaviors they deem unprofessional of themselves. Both theoretical perspectives support the hypothesis that officers wearing BWCs will initiate fewer encounters with the public than officers without cameras. As Braga et al. (2018b, p. 517) concluded regarding BWC-related reductions in complaints against officers, a possible outcome of self-initiated patrolling, “it remains unclear whether deterrence, self-awareness, or both are generating the observed effects.”

The two perspectives are especially relevant to de-policing because the current literature on officer perceptions of camera use highlights concerns about external sanctions and internal values conflicts. For example, officers in Katz et al.’s study of BWCs in Phoenix voiced apprehension that footage might be used against them after being reviewed by supervisors or the public (Katz et al., 2015). Newell and Greidanus’s (2018) officer surveys documented concern that BWC footage would be very or somewhat likely to lead to increased disciplinary measures against line officers. Such concerns are not found only among patrol officers, however. In separate focus groups conducted with rank-and-file and supervisors as part of the research reported here, participants in both groups worried that BWC videos would increase distrust between the ranks because of fears that minor infractions would result in disciplinary actions. A similar finding was noted in a study conducted by Miller and Toliver (2014).

During the pilot study, the department had two policies related to the salience of the deterrence and self-awareness theories. First, the officers assigned BWCs were required to turn on their cameras during any official police encounter with the public, with a few well-defined exceptions such as cases involving medical issues or minors in a school setting. Second, senior officers were required to review randomly selected portions of BWC footage of the officers they supervised and to flag behaviors possibly violating laws or departmental policies. Patrol officers were aware of both requirements, detailed in training and documents they were required to read, so their behavior during the study was likely shaped by both social psychological forces.

The results of studies exploring whether these external or internal pressures lead to behavioral changes in police are mixed (c.f. Lum et al., 2019, 2020). The largest body of work, especially in the earliest BWC studies, focused on changes in the number of use of force complaints against officers after BWCs were deployed. Many of these studies found that BWC-equipped officers received significantly fewer such complaints than their non-camera wearing colleagues. The magnitude of the difference varied widely across studies, ranging from a 26% to a 59% overall reduction (Ariel et al., 2015; Braga et al., 2018b; Jennings et al., 2014; White et al., 2017), although some studies found no differences (Ariel, 2016; Edmonton Police Service, 2015; Grossmith et al., 2015; Yokum et al., 2019).

Recent research dealing with the effects of BWCs on pedestrian stops has produced more mixed results. Ready and Young (2015) noted that camera wearing officers in Mesa, AZ, were less likely to stop and frisk, a legally specific form of pedestrian stop, but were more likely to initiate contacts with the public and issue citations. Unlike other studies, their data source was field contact reports created by the officer after the stop occurred, rather than computer-aided dispatch (CAD) records generated on-scene. Groff and her research team (2020) also used Terry-like pedestrian stops as their measure of de-policing and noted a decrease of 46% during their study of the Philadelphia Police Department.² Likewise, Huff and associates (2020) found that self-initiated contacts, defined as subject/vehicle stops, by officers in Phoenix, AZ, declined once officers activated their cameras. In their study of the Milwaukee Police Department, Peterson et al. (2018) also reported fewer subject stops (detention of a person for questioning) by officers wearing a BWC.

Along with those studies reporting a reduction in self-initiated subject stops by BWC-wearing officers, there are studies which report no difference in the levels of this self-initiated activity by the presence or absence of BWCs. Hughes, Campbell, and Schaefer (2020) found no significant differences in overall self-initiated activity among Louisville, KY, Metro Police officers, but did find a significant reduction of such activity in socially disadvantaged areas of the city. Braga et al. (2017) also found no significant differences by BWC status in self-initiated activity among officers in Las Vegas. Their measure was a monthly count of responses to dispatched, self-initiated and crime report events. A subsequent study based in the same department found no significant differences in officer-initiated stops by whether a BWC was present or not (Braga et al., 2018b). Another study by Braga and associates (2020) in Boston also found no significant differences in broadly defined self-initiated activity.

In contrast, Zamoff et al. (2021) found a significant increase in police self-initiated non-traffic stops by BWC-equipped officers. They used New York Police Department’s annually published Stop, Question, and Frisk dataset to evaluate the level of self-initiated activity, which was defined as a subject stop based upon suspected criminal behavior. Similarly, Headley et al. (2017) found that the incidence of self-initiated, unspecified subject stops by BWC-wearing officers increased in Hallandale Beach, FL, but their sample was not large enough to support reliable conclusions.

A unique finding on the effects of BWCs was reported by Wallace et al. (2018) in their study of the Spokane, WA, Police Department. Although they hypothesized that BWCs would increase officer passivity, the authors found no difference in the prevalence of self-initiated calls between treatment and control groups, while noting that calls declined in both groups. However, when the interactions between group assignment and intervention timeframes were further analyzed, the authors found that officers in the treatment group had higher odds of engaging in a self-initiated activity than their control counterparts, during both the experimental and post-experimental periods.

A follow-up study in Milwaukee, WI, by Lawrence and Peterson (2020) found no difference in the overall levels of self-initiated police activities between treatment and control groups but when the types of activities they initiated were separately examined, differences emerged. The researchers divided non-traffic, self-initiated police actions into three categories: subject stops, business checks and park and walks. They found a significant reduction in subject stops by officers wearing cameras (approximately 8%), no change in business checks and a significant increase (approximately 23%) in park and walks. They attributed these differences to the nature of the self-initiated activity: subject stops offer higher risks of confrontations whereas park and walks are more community policing-oriented with less chance of confrontations. The study’s business check and park and walk measures are similar to our measures of non-traffic, self-initiated police activities which also includes subject stops.

In summary, prior research does not present a consistent picture of the effects of BWCs on self-initiated police encounters with the public stemming from foot patrols. The earlier studies have largely focused on the specific outcomes of such encounters rather than the general category of discretionary patrol work. For example, an officer’s single decision to engage in proactive foot patrol may result in three stop and frisk encounters, two subject stops or nothing measurable. Since our study focuses on the number of foot patrols conducted, regardless of outcomes, the prior research offers little empirical guidance.

There is, however, overwhelming evidence that BWCs do not reduce the number of self-initiated traffic stops. As opposed to foot patrols, the majority of studies that looked at traffic stops measure the self-initiated activity itself and not its outcomes. Pearson and associates (2018) found in Milwaukee, WI, that the incidence of self-initiated traffic stops did not differ between treatment and control groups. A follow-up study in Milwaukee by Lawrence and Peterson (2020) also found no difference in the levels of self-initiated traffic stops between the two groups although there were differences based on the race and gender of the officers. Likewise, Yokum et al. (2019) found that cameras did not affect the number of traffic stops in Washington, DC, and Groff and associates (2020) found no significant BWC effects on the number of traffic stops made by Philadelphia police officers.

This study on the effects of BWCs on de-policing takes a different approach by utilizing inclusive measures of self-initiated actions. Rather than focusing only on outcomes that originate from self-initiated behavior, such as pedestrian stops, stop and frisks, and arrests, this study measures every action an officer took after the decision to leave a patrol car was made, but not in response to a dispatched call for service. As prior research shows, foot patrols involve much more than enforcement activities such as stop and frisks and arrests. Piza (2018) divides foot patrols into Enforcement and Guardian Actions and reports that Guardian Actions are approximately three times more frequent than Enforcement Actions during foot patrols conducted in Newark, NJ. Guardian Actions include citizen contacts, business checks, bus checks and taxi inspections while Enforcement Actions involve arrests, field interrogations and quality of life summons. By using an inclusive approach, the measure of foot patrol employed here combines all such actions an officer can undertake while on foot patrol and the measure of traffic stops is similarly defined.³

The research design of this study expands our knowledge of self-initiated police activity in four ways. First, it measures of the number of times officers initiated foot and traffic patrols rather than what the officers did during those patrols. Prior studies have focused on a variety of patrol outcomes which has made drawing firm conclusions about the effects of BWCs difficult. Second, it was conducted in collaboration with the FCPD, a large, urban and progressive police department accredited by the Commission on Accreditation for Law Enforcement Agencies (CALEA) which indicates a highly professional organization. Although prior research has shown inconsistent support for the deterrence or self-awareness perspectives, we test them again as predictors of a decline in two types of self-initiated police activity among well-trained officers wearing BWCs compared with colleagues not equipped with cameras. Third, although this study employs a quasi-experimental design, the unusual scheduling of squads as detailed in the methods section substantially reduced the likelihood that officers with cameras and those without them were present together at any public encounter. In other words, there was little if any contamination effects between the treatment and control groups. Fourth, we used Stata’s itsa program for interrupted time series data, an analysis tool that is more sensitive to variation in data points and produces more robust estimators than the analytical techniques used in past studies (Linden, 2015).

Methods

Design

This study is part of a larger project for which researchers collected qualitative and quantitative data using numerous methods and that addressed multiple BWC-related issues.⁴ It employed a quasi-experimental design where three of the department’s eight police districts were selected to participate in the experiment. Of the three police districts selected, two were chosen by the department (Districts 1 and 2) because of their high volume of calls for service involving violent and property crime and the greater diversity of their residents. The third, District 3, was selected by the researchers because it was more representative of the five, non-selected districts. The top portion of Table 1 presents demographic and crime data on the three districts and their residents. Nearly 12% and 10% of the District 1 and 2 residents, respectively, lived below the US poverty line compared with approximately 4% of the District 3 residents. The percentages of minority residents ranged from a high of 70% in District 1 to a low of 40% in District 3. The districts also had the same high-to-low ordering in their rates of property and violent crime, with District 2 having the highest rate of the three.

Table 1.

District and Officer Characteristics.

District characteristics	District 1	District 2	District 3
% Below the Poverty Line	11.8	9.6	4.4
% Minority	70	55	40
Property crime rate per 100,000	2040	2869	1326
Violent crime rate per 100,000	790	1232	458
Officers’ characteristics
Years of service (mean)	10.5	11.4	14.6
Rank
Patrol officer	88%	85%	85%
Supervisor, above	12%	15%	15%
Gender
Male	85%	77%	94%
Female, other	15%	23%	6%
Education
Associates degree, less	35%	35%	39%
Four year degree, more	65%	65%	61%
Ethnicity
White	76%	68%	82%
Black, other	24%	32%	18%
District staffing
Squad B/Treatment group (162 officers)	58	58	46
Squad A/Control group (134 officers)	51	41	42

The lower portion of Table 1 presents the demographic characteristics of the officers serving in the three districts. The officers assigned to Districts 1 and 2 had similar years of service (10.5 and 11.4, respectively) but had fewer years when compared to District 3 (14.6). The districts were nearly identical in the proportion holding the front-line rank of patrol officer. Differences were found on gender among the three districts: District 2 had the highest percentage of female officers (23%), followed by District 1 (15%) and then District 3 (6%). Table 1 also shows that the level of education was very similar across the three districts. Finally, the racial diversity among officers in the three districts showed, like gender, that District 2 had the highest level (32% minority), followed by District 1 (24% minority) and District 3 (18% minority).

Prior to the evaluation, patrol officers in each district were routinely divided into two squads designated as A or B. The squads worked different days of the week over a 2-week pay period, resulting in very little interaction between them.⁵ The researchers flipped a coin and Squad B was assigned to wear the cameras for the 6-month pilot period as members of the treatment group. Although there were differences in officer demographics across the three districts, there were no statistically significant differences in the officers’ characteristics between the treatment (Squad B) and control (Squad A) groups at the beginning of the pilot period.⁶ The bottom portion of Table 1 presents data on staffing by district and by squad and shows the number of officers available for patrol, excluding those who were on light duty because of injuries. Squad B also included six to eight Neighborhood Patrol Unit officers in Districts 1 and 2 who patrolled specific areas on bicycle.

Unlike most of the randomized controlled trial (RCT) designs utilized in previous BWC studies, the squad schedule offered a unique way to minimize contamination effects stemming from both Squad A and Squad B officers responding to the same call. As noted by Lum et al. (2020), a similar design employed by Bennett et al. (2019) reduced the chances for both within-subject and across-subject contamination.⁷ Indeed, the only time that a member of one squad worked with an officer from the other was when a squad had an insufficient number of officers for effective patrol (e.g., due to sick or vacation leave). A volunteer officer from the other squad was paid overtime to work with the understaffed squad. However, policy dictated that if a Squad B member worked with Squad A, they did not wear their camera while doing so. When a Squad A officer worked on a Squad B shift, the Squad A officer was not issued a BWC.⁸

Measurement

The independent variable, Presence of Camera, is a dichotomous variable based upon whether the officer was assigned a camera (i.e., in Squad B, the treatment group) or not (in Squad A, the control group). The data were collected for 52 weeks prior to the pilot, 26 weeks during the pilot and for 18 weeks after the pilot ended, for a total of 96 weeks of data collection.⁹

Data measuring both dependent variables were retrieved from the department’s electronic data storage system. The first dependent variable, Self-initiated Foot Patrols, is defined as any time an officer leaves their car to engage in proactive, non-dispatched foot patrol. This includes such activities as walking a beat, checking out a building or perusing area malls and businesses. The data were collected from the department’s Integrated Law Enforcement Automated Data System (ILEADS) module. This module recorded the time and location of officers when they “marked-out,” i.e., they entered a specific code into their onboard computer before leaving the patrol car to engage in a self-initiated foot patrol. Since FCPD deploys single-officer patrol cars, high levels of officer compliance with this protocol are likely because it is well-understood to protect officers in situations where they may encounter a threat while out of their patrol car. The data showed that Squad B officers engaged in 10,322 self-initiated foot patrols while Squad A officers engaged in 9,629 of them over the 96 weeks.

Data on the second dependent variable, Self-initiated Traffic Stops, was also collected from the ILEADS module. It recorded the time and location of officers when they “marked out” by entering a different code into their computer indicating they were leaving the patrol car to make a non-dispatched traffic stop for a driving violation or suspicious vehicle activity. Over 96 weeks, Squad B made 993 self-initiated traffic stops and Squad A made 1,003 such stops.

Analyses

The data analyses were conducted using an interrupted times series statistical program, Stata’s itsa, because the exact dates were known when the officers in the three districts were assigned BWCs and when the cameras were removed. Stata’s itsa uses robust estimators that can handle autocorrelation and possible heteroskedasticity and was selected over an ARIMA-based program because of its flexibility and applicability to time series interpretations (Linden, 2015). The itsa procedure allows for multiple time periods, including before BWCs were deployed, once they were assigned and the period following their removal (Linden, 2015). The itsa procedure also allows for the inclusion of a control group. Preliminary analyses indicated that the treatment and control groups were balanced, i.e., showed similar numbers of each of the two types of stops during the pre-intervention period, so a two-lag model was used. We also used weekly differencing to account for the fact that the squads rotated their scheduled workdays each week.¹⁰ We then used the point estimates for linear combinations of the coefficients as produced by itsa as input for the lincom commands in Stata to look for statistical significance in the trends of foot patrol stops between the treatment and control groups across the three time periods. We did the same for traffic stops and also for the difference between the two types of patrols, per Linden (2017).

Findings

The findings regarding each self-initiated activity are presented in two forms. First, Figures 1 and 2 show the spread around the actual and predicted slopes for both the treatment (Squad B) and control (Squad A) groups. Second, three sets of tables present the t-statistics indicating the magnitude of differences between the number of self-initiated activities conducted during the pre-intervention period (Weeks 1–53, shown in Tables 2 and 5), the intervention period (Weeks 54–79, shown in Tables 3 and 6) and the post-intervention period (Weeks 80–94, shown in Tables 4 and 7). These comparisons were made because officer behavior may have changed significantly between the first two periods (i.e., the pre-intervention and intervention), between the second and third periods (the intervention and post-intervention), in neither comparison or in both.

Figure 1.

Interrupted time series analysis of BWCs on self-initiated foot patrol.

Figure 2.

Interrupted time series analysis of BWCs on self-initiated traffic stops.

Table 2.

Comparisons of Linear Pre-intervention Trends on Foot Patrol: Week 1–53.

Linear trend	Coefficient	SE	t	p-value	95% Confidence interval
Treatment	.068	.320	.210	.830	−.550	.697
Control	−.190	.278	−.680	.495	−.739	.359
Difference	−.259	.423	−.610	.541	−1.093	.576

Table 3.

Comparisons of Linear Intervention Trends on Foot Patrol: Week 54–79.

Linear trend	Coefficient	SE	t	p-value	95% Confidence interval
Treatment	.135	.570	.236	.814	−.990	1.259
Control	−.159	.557	−.286	.775	−1.258	.939
Difference	.294	.796	.369	.713	−1.278	1.866

Table 4.

Comparisons of Linear Post-intervention Trends on Foot Patrol: Week 80–94.

Linear trend	Coefficient	SE	t	p-value	95% Confidence interval
Treatment	.654	1.120	.584	.560	−1.556	2.865
Control	.039	.773	.071	.960	−1.487	1.565
Difference	.615	1.361	.452	.652	−2.071	3.301

Figure 1 and supporting Tables 2, 3, and 4 present the results on Self-initiated Foot Patrols. It is apparent from Figure 1 that there is no difference in the number of foot patrols the treatment and control groups made prior to, during, and after the pilot program ended. This visual determination is statistically supported in Tables 2, 3, and 4. The difference coefficient for the pre-intervention period (see Table 2) had a t value of −.610, p = .541. This indicates that there are no statistically significant differences between groups prior to implementing the BWC program. Similarly, the largest t value in the intervention period (see Table 3) was the difference coefficient of .369, p = .713 which is also not statistically significant. Finally, the t value for the difference between groups in the post-intervention period (see Table 4) was .452, p = .652, again indicating that there is no significant difference after the BWCs were removed. These statistics support the conclusion that the presence of BWCs did not influence the number of Self-initiated Foot Patrols during any of the three periods analyzed.

Figure 2 and supporting Tables 5, 6, and 7 present the interrupted time-series findings for Self-initiated Traffic Stops. As clearly shown in Figure 2, there is no difference between the treatment and control groups in the number of traffic stops they made prior to, during, and after the pilot program ended. This visual determination is statistically supported in Tables 5, 6, and 7.

Table 5.

Comparisons of Linear Pre-intervention Trends on Traffic Stops: Week 1–53.

Linear trend	Coefficient	SE	t	p-value	95% Confidence interval
Treatment	.694	1.383	.500	.616	−2.035	3.422
Control	−.214	.931	−.230	.819	−2.050	1.623
Difference	−.908	1.667	−.540	.587	−4.197	2.382

Table 6.

Comparisons of Linear Intervention Trends on Traffic Stops: Week 54–79.

Linear trend	Coefficient	SE	t	p-value	95% Confidence interval
Treatment	1.128	2.733	.413	.680	−4.266	6.522
Control	−.449	2.808	−.160	.873	−5.989	5.092
Difference	1.576	3.918	.402	.688	−6.157	9.309

Table 7.

Comparisons of Linear Post-intervention Trends on Traffic Stops: Week 80–94.

Linear trend	Coefficient	SE	t	p-value	95% Confidence interval
Treatment	−.154	3.564	−.430	.966	−7.189	6.880
Control	1.770	6.114	.290	.773	−10.296	13.835
Difference	−1.924	7.077	−.272	.786	−15.89	12.042

The largest t value for the pre-intervention period (see Table 5) was for the difference coefficient between Squads A and B (t = −.540, p = .578), which is not significant and indicates that the number of traffic stops during pre-intervention did not vary by squad. The same is found for the intervention period (see Table 6) with the difference coefficient having a t value of .402, p = .688, indicating that there are no significant differences between the groups regarding the number of traffic stops. Finally, for the post-intervention period (see Table 7), the difference coefficient between the squads had a t value of −.272, p = .786, denoting no significant difference between them. These findings support the conclusion that the wearing of BWCs did not affect the number of Self-initiated Traffic Stops officers made during any of the three periods analyzed. Further analyses were undertaken to determine if the overall findings differed by police district, and they did not. There were no significant differences across the three districts in the number of either foot patrol or traffic patrols conducted by the treatment and control groups.¹¹

Collectively, the findings indicate that the presence of BWCs did not influence the number of self-initiated foot patrols or traffic stops conducted by officers before, during or after the pilot program. Instead, the levels of policing before cameras were deployed remained nearly constant after BWCs were assigned and continued after the pilot program was terminated.

Conclusions

This study seeks to extend knowledge about the effects of BWCs on police work and to test hypotheses that deployment of the cameras leads to de-policing. Measures of the two main types of self-initiated police behavior, foot patrols and traffic stops, were collected during a lengthy pilot program based on a quasi-experimental design. The data were collected in three large, diverse districts of an urban police organization and analyzed using the powerful Stata itsa statistical program, well-suited to the type of data collected and the study’s extended timeline. Like some prior de-policing studies that used a variety of research designs, measures and data collection protocols, this study found no significant differences over three time periods in the number of self-initiated foot patrols or traffic stops made by officers assigned cameras compared to officers that were not. These null findings support the conclusion that police officers do not reduce their likelihood of stopping drivers or patrolling neighborhoods on foot when they wear body-worn cameras.

Limitations

Like most social science research, this study has three limitations that may affect the validity of the findings. First, the study employed a quasi-experimental design where squads of officers, not individual officers, were randomly assigned BWCs or not. Although we found no significant differences between treatment and control groups on five officer characteristics and the demographics of the districts they patrolled, there may be effects from unknown and unmeasured variables, such as differences in a squad’s proclivity to take self-initiated actions when on patrol. Even though the unique squad and shift policies of the department likely reduced contamination effects commonly found in other studies with treatment and control groups, some contamination may have occurred.

Second, measurement issues might also cast doubt on the validity of the findings. Measures of both self-initiated foot patrols and traffic stops assumed an officer marked-out before engaging in either activity. Human error can rarely be ruled out in a behavioral study and it is plausible that officers occasionally forgot to do so. There is no reason to believe, however, that this forgetfulness was more pervasive among one squad than the other or among officers preparing to do a foot patrol or traffic stop. The safety rationale for the mark-out requirement likely produced high levels of compliance among all officer groups.

Third, and germane to most social science studies, is the issue of generalizability. This study was conducted in only three of the FCPD’s eight districts (two selected by the agency and one selected by the researchers that typified the remaining non-selected districts) and it is possible that officers' behavior in the other five districts might be different, so these results cannot be generalized to them. Generalizing to all police departments in the same state, same region, or in the US is even more problematic.

Discussion and Policy Implications

The findings from this study extend the literature on the effects of BWCs on the policing activities of patrol officers. By using data collected over a long timeframe and sophisticated analysis techniques, the study shows that officers wearing cameras choose to engage in similar numbers of self-initiated foot patrol stops and traffic stops as officers without them, i.e., the presence of a camera does not lead to de-policing. These finding also suggests that deterrence and self-awareness theories have little predictive value regarding the behavior of BWC-wearing officers.

Based upon the limitations of this study, additional research on BWC-related de-policing should be conducted in departments that vary in characteristics not explored here, such as strength of community relationships, the size of their marginalized communities, the proportion of youth within their population, the number and type of calls for service, the level and types of crime problems, the records system used and how it might provide better measurements, and the BWC policies and practices in place. These additional factors should be investigated to understand more fully their impact on the behavior of officers and members of the public. Finally, further research related to officer decision-making, self-initiated actions and patrol methods with and without BWCs would provide valuable additional insights.

The major policy implication arising from the study is that the deployment of BWCs does not appear to decrease or increase the self-initiated behaviors of officers on foot or traffic patrol. Police officials, officers themselves and community members may have feared such an impact but this study challenges claims of de-policing.

The decision whether to adopt BWCs is not, for most police departments, a simple matter. Chiefs and local officials must weigh three factors: the likely organizational-financial benefits such as faster resolution of citizen complaints and use of force incidents (Braga et al, 2017; Williams, et al., 2021), and officer oversight; purely financial costs (e.g., significant monetary and personnel costs associated with the purchase, maintenance and data storage), and the level of public safety that BWC-equipped officers provide. This study addresses the third factor, likely the most important to community leaders, members and policy makers, and indicates that the public safety produced by officers on patrol is not impacted by their wearing BWCs.

Footnotes

Acknowledgments

The authors would like to thank the administration and officers of the Fairfax County Police Department for their support of this research.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Charles Koch Foundation and School of Public Affairs, American University.

Author’s Note

An earlier version of this manuscript was presented at the annual meetings of the American Society of Criminology in San Francisco.

ORCID iDs

Richard R. Bennett

Brad Bartholomew

Notes

Author Biographies

Richard R. Bennett, PhD, is a Professor of Justice in the Department of Justice, Law and Criminology at American University. He has conducted research and published a book and over 65 articles in the areas of police attitudes and behaviors as well as comparative crime and criminal justice. Earlier he served as a police officer, a police trainer, criminal investigator, and is currently a criminal justice consultant to national and international governmental commissions, universities, and contract research organizations. He has been both a Senior Fulbright Scholar and president of the Academy of Criminal Justice Sciences. He has won several national awards for his research and work in the field.

Brad Bartholomew, PhD is a Senior Professorial Lecturer in the Department of Justice, Law and Criminology at American University in Washington, DC. He earned his PhD in Criminal Justice and Criminology from The University of Maryland. Dr. Bartholomew has worked with local, state and federal agencies to promote professional standards and accountability in policing. His primary research interests include policing and police reform, with a focus on technology and police-citizen interactions.

Sandra Baxter, PhD, is a Faculty Fellow in Residence in the Department of Justice, Law and Criminology at American University in Washington, DC. She has published numerous articles and three books on a wide range of topics including US. Civil Service reform, HIV/AIDS education and prevention, and women and politics. She is a graduate of the Universities of Chicago and Michigan.

Holly Champagne is currently a PhD candidate in the Department of Justice, Law and Criminology at American University in Washington, DC. She holds a Juris Doctorate, with a special certificate in Criminal Law, from the DePaul University College of Law, a Master of Science in Criminal Justice from the University of Cincinnati. She gained practical criminal justice experience working as a civilian within police departments in both intelligence and police administration and as a student prosecutor within two separate state’s attorney offices.

Eric R. Schuler, PhD, is a quantitative/computational research methodologist at American University’s Center for Teaching, Research, & Learning. He is an experimental psychologist by training. His past research investigated correlates of posttraumatic stress disorder symptomology and the impact of potentially traumatic events on perceptions of the self, world, and others. His current research focuses on psychometrics, specifically the usage of Bayesian methods with small sample sizes.

References

Ariel

(2016). Increasing cooperation with the police using body worn cameras. Police Quarterly, 19(3), 326–362. https://doi.org/10.1177/1098611116653723

Ariel

Farrar

W. A.

Sutherland

(2015). The effect of police body-worn cameras on use of force and citizens’ complaints against the police: A randomized controlled trial. Journal of Quantitative Criminology, 3(31), 509–535. https://doi.org/10.1007/s10940-014-9236-3

Bennett

R. R.

Bartholomew

Champagne

(2019). Fairfax County Police Department’s body-worn camera pilot project: An evaluation. Fairfax County Police Department. https://www.fairfaxcounty.gov/police/sites/police/files/assets/documents/fcpdfinalreport06_25_19.pdf

Braga

Coldren

Sousa

Rodriguez

Alper

(2017). The benefits of body-worn cameras: New findings from a randomized controlled trial at the Las Vegas Metropolitan police department. CNA Analysis & Solutions. https://www.ojp.gov/pdffiles1/nij/grants/251416.pdf

Braga

A. A.

Barao

McDevitt

Zimmerman

Evans

P. C. W.

Buckley

S. K.

(2018a). The impact of body-worn cameras on complaints against officers and officer use of force incident reports: Preliminary evaluation findings. School of Criminology and Criminal Justice, Northeastern University. https://www.bwctta.com/sites/default/files/Files/Resources/BPD%2BBWC%2BRCT%2Bpreliminary%2Bimpact%2Breport.pdf

Braga

A. A.

Barao

L. M.

Zimmerman

G. M.

Douglas

Sheppard

(2020). Measuring the direct and spillover effects of body worn cameras on the civility of police–citizen encounters and police work activities. Journal of Quantitative Criminology, 36(4), 851–876. https://doi.org/10.1007/s10940-019-09434-9

Braga

A. A.

Sousa

W. H.

Coldren

J. R.

Jr. Rodriguez

(2018b). The effects of body-worn cameras on police activity and police-citizen encounters: A randomized controlled trial. Journal of Criminal Law and Criminology, 108(3), 511–538. https://scholarlycommons.law.northwestern.edu/jclc/vol108/iss3/3

Crow

M. S.

Snyder

J. A.

Crichlow

V. J.

Smykla

J. O.

(2017). Community perceptions of police body-worn cameras: The impact of views on fairness, fear, performance, and privacy. Criminal Justice and Behavior, 44(4), 589–610. https://doi.org/10.1177/0093854816688037

Davenport

V. E.

(2017). Commentary: Investigative police stops—necessary or insidious? A practitioner’s viewpoint. Public Administration Review, 77(2), 179–180. https://doi.org/10.1111/puar.12741

10.

Edmonton Police Service (2015). Body-worn video: Considering the evidence: Final report of the Edmonton Police Service body-worn video pilot project. Edmonton Police Service. https://bwvsg.com/wp-content/uploads/2015/06/Edmonton-Police-BWV-Final-Report.pdf

11.

Fredericks

(2021). Video & policing 9/11 to today: The road to video literacy for investigators. Police Chief, 88(9), 102–106.

12.

Goodison

Wilson

(2017). Citizen perceptions of body-worn cameras: A randomized controlled trial. Police Executive Research Forum. https://perf.memberclicks.net/assets/bodyworncameraperceptions.pdf

13.

Groff

E. R.

Haberman

Wood

J. D.

(2020). The effects of body-worn cameras on police-citizen encounters and police activity: Evaluation of a pilot implementation in Philadelphia, PA. Journal of Experimental Criminology, 16(4), 463–480. https://doi.org/10.1007/s11292-019-09383-0

14.

Grossmith

Owens

Finn

Mann

Davies

Baika

(2015). Police, camera, evidence: London’s cluster randomised controlled trial of body worn video. College of Policing Limited and the Mayor’s Office for Policing and Crime (MOPAC). https://whatworks.college.police.uk/Research/Documents/Police_Camera_Evidence.pdf

15.

Headley

A. M.

Guerette

R. T.

Shariati

(2017). A field experiment of the impact of body-worn cameras (BWCs) on police officer behavior and perceptions. Journal of Criminal Justice, 53(C), 102–109. https://doi.org/10.1016/j.jcrimjus.2017.10.003

16.

Hedberg

E. C.

Katz

C. M.

Choate

D. E.

(2017). Body-worn cameras and citizen interactions with police officers: Estimating plausible effects given varying compliance levels. Justice Quarterly, 34(4), 627–651. https://doi.org/10.1080/07418825.2016.1198825

17.

Huff

Katz

C. M.

Hedberg

E. C.

(2020). A randomized controlled trial of the impact of body-worn camera activation on the outcomes of individual incidents. Journal of Experimental Criminology, 18(2), 247–272. https://doi.org/10.1007/s11292-020-09448-5

18.

Hughes

T. W.

Campbell

B. A.

Schaefer

B. P.

(2020). The influence of body-worn cameras, minority threat, and place on police activity. Journal of Community Psychology, 48(1), 68–85. https://doi.org/10.1002/jcop.22299

19.

Hyland

S. S.

(2018). Body-worn cameras in law enforcement agencies, 2016. U.S. Department of Justice, Once of Justice Programs, Bureau of Justice Statistics.

20.

Jennings

W. G.

Lynch

M. D.

Fridell

L. A.

(2014). Cops and cameras: Officer perceptions of the use of body-worn cameras in law enforcement. Journal of Criminal Justice, 42(6), 549–556. https://doi.org/10.1016/j.jcrimjus.2014.09.008

21.

Jennings

W. G.

Lynch

M. D.

Fridell

L. A.

(2015). Evaluating the impact of police officer body-worn cameras (BWCs) on response-to-resistance and serious external complaints: Evidence from the orlando police department (OPD) experience utilizing a randomized controlled experiment. Journal of Criminal Justice, 43(6), 480–486. https://doi.org/10.1016/j.jcrimjus.2015.10.003

22.

Katz

C. M.

Choate

D. E.

Ready

J. R.

Nuño

(2014). Evaluating the impact of officer worn body cameras in the Phoenix Police Department. Center for Violence Prevention & Community Safety, Arizona State University. https://publicservice.asu.edu/sites/default/files/ppd_spi_feb_20_2015_final.pdf

23.

Katz

C. M.

Kurtenbach

Choate

D. E.

White

M. D.

(2015). Smart Policing Initiative: Evaluating the impact of police officer body-worn cameras. CNA Analysis & Solutions. https://www.strategiesforpolicinginnovation.com/sites/default/files/spotlights/PhoenixSPISpotlightFINAL.pdf

24.

Lawrence

D. S.

Peterson

B. E.

(2020). How do body-worn cameras affect the amount and makeup of police-initiated activities? A randomized controlled trial in milwaukee, WI. Journal of Experimental Criminology, 16(4), 481–503. https://doi.org/10.1007/s11292-019-09385-y

25.

Linden

(2015). Conducting interrupted time-series analysis for single- and multiple-group comparisons. The Stata Journal, 15(2), 480–500. https://doi.org/10.1177/1536867X1501500208

26.

Linden

(2017). A comprehensive set of postestimation measures to enrich interrupted time-series analysis. The Stata Journal, 17(1), 73–88. https://doi.org/10.1177%2F1536867X1701700105

27.

Lum

Koper

C. S.

Wilson

D. B.

Stoltz

Goodier

Eggins

Higginson

Mazerolle

(2020). Body-worn cameras’ effects on police officers and citizen behavior: A systematic review. Campbell Systematic Reviews, 16(3), e1112. https://doi.org/10.1002/cl2.1112

28.

Lum

Stoltz

Koper

C. S.

Scherer

J. A.

(2019). Research on body-worn cameras: What we know, what we need to know. Criminology & Public Policy, 18(1), 93–118. https://doi.org/10.1111/1745-9133.12412

29.

Lynch

M. D.

(2015). Police technology. In Jennings

(Ed.), The encyclopedia of crime & punishment (pp. 1–5). https://doi.org/10.1002/9781118519639.wbecpx283

30.

MacDonald

J. M.

Klick

Grunwald

(2016). The effect of private police on crime: Evidence from a geographic regression discontinuity design. Journal of the Royal Statistical Society. Series A (Statistics in Society), 179(3), 831–846. https://doi.org/10.1111/rssa.12142

31.

Maskaly

Donner

Jennings

W. G.

Ariel

Sutherland

(2017). The effects of body-worn cameras (BWCs) on police and citizen outcomes: A state-of-the-art review. Policing: An International Journal of Police Strategies & Management, 40(4), 672–688. https://doi.org/10.1108/PIJPSM-03-2017-0032

32.

Miller

Toliver

(2014). Implementing a body-worn camera program: Recommendations and lessons learned. Police Executive Research Forum. https://www.justice.gov/iso/opa/resources/472014912134715246869.pdf

33.

Newell

B. C.

Greidanus

(2018). Officer discretion and the choice to record: Officer attitudes towards body-worn camera activation. North Carolina Law Review, 96(5), 1525–1578. https://scholarship.law.unc.edu/nclr/vol96/iss5/8

34.

Nowacki

J. S.

Willits

(2018). Adoption of body cameras by United States police agencies: An organisational analysis. Policing and Society, 28(7), 841–853. https://doi.org/10.1080/10439463.2016.1267175

35.

Pearson

Rowe

Turner

(2018). Policy, practicalities, and PACE s. 24: The subsuming of the necessity criteria in arrest decision making by frontline police officers. Journal of Law and Society, 45(2), 282–308. https://doi.org/10.1111/jols.12087

36.

Peterson

B. E.

La Vigne

Lawrence

D. S.

(2018). The Milwaukee Police Department’s body-worn camera program. Urban Institute. https://www.urban.org/sites/default/files/publication/98461/the_milwaukee_police_departments_body_worn_camera_program_2.pdf

37.

Piza

E. L.

(2018). The effect of various police enforcement actions on violent crime: Evidence from saturation foot-patrol intervention. Criminal Justice Policy Review, 29(6–7), 611–629. https://doi.org/10.1177/0887403417725370

38.

President’s Task Force on 21st Century Policing (2015). Final report of the President’s task force on 21st century policing. Office of Community Oriented Policing Services. https://cops.usdoj.gov/pdf/taskforce/taskforce_finalreport.pdf

39.

Ready

J. T.

Young

J. T.

(2015). The impact of on-officer video cameras on police–citizen contacts: Findings from a controlled experiment in Mesa, AZ. Journal of Experimental Criminology, 11(3), 445–458. https://doi.org/10.1007/s11292-015-9237-8

40.

Reaves

B. A.

(2015). Local police departments, 2013: Personnel, policies, and practices. Bureau of Justice Statistics. https://bjs.ojp.gov/content/pub/pdf/lpd13ppp.pdf

41.

Rosenfeld

(2016). Documenting and explaining the 2015 homicide rise: Research directions. National Institute of Justice. https://doi.org/10.1177/00111287211039994

42.

Rushin

Edwards

(2016). De-policing. Cornell Law Review, 102(3), 721–782. http://scholarship.law.cornell.edu/clr/vol102/iss3/3

43.

Sousa

W. H.

Miethe

T. D.

Sakiyama

(2018). Inconsistencies in public opinion of body-worn cameras on police: Transparency, trust, and improved police–citizen relationships. Policing: A Journal of Policy and Practice, 12(1), 100–108. https://doi.org/10.1093/police/pax015

44.

Terry v. Ohio, 392 U.S. 1 (1968). https://www.oyez.org/cases/1967/67

45.

Wallace

White

M. D.

Gaub

J. E.

Todak

(2018). Body-worn cameras as a potential source of depolicing: Testing for camera-induced passivity. Criminology, 56(3), 481–509. https://doi.org/10.1111/1745-9125.12179

46.

White

M. D.

(2014). Police officer body-worn cameras: Assessing the evidence. Office of Community Oriented Policing Services.

47.

White

M. D.

Todak

Gaub

J. E.

(2017). Assessing citizen perceptions of body-worn cameras after encounters with police. Policing: An International Journal of Police Strategies and Management, 40(4), 689–703. https://doi.org/10.1108/PIJPSM-07-2016-0105

48.

Williams

M. C.

Jr. Weil

Rasich

E. A.

Ludwig

Chang

Egrari

(2021). Body-worn cameras in policing: Benefits and costs. National Bureau of Economic Research. https://doi.org/10.3386/w28622

49.

Yokum

Ravishankar

Coppock

(2019). Evaluating the effects of police body-warn cameras: A randomized controlled trial. PNAS, 116(21), 10329-10332. https://doi.org/10.1073/pnas.1814773116.

50.

Zamoff

M. E.

Greenwood

B. N.

Burtch

(2021). Who watches the watchman: Evidence of the effect of body-worn cameras on New York City policing. The Journal od Law, Economics, and Organization, 3(1), 161-195.