Abstract
Background:
Recent technological advances have much potential for improving police performance, but there has been little research testing whether they have made police more effective in reducing crime.
Objective:
To study the uses and crime control impacts of mobile computing technology in the context of geographically focused “hot spots” patrols.
Research Design:
An experiment was conducted using 18 crime hot spots in a suburban jurisdiction. Nine of these locations were randomly selected to receive additional patrols over 11 weeks. Researchers studied officers’ use of mobile information technology (IT) during the patrols using activity logs and interviews. Nonrandomized subgroup and multivariate analyses were employed to determine if and how the effects of the patrols varied based on these patterns.
Results:
Officers used mobile computing technology primarily for surveillance and enforcement (e.g., checking automobile license plates and running checks on people during traffic stops and field interviews), and they noted both advantages and disadvantages to its use. Officers did not often use technology for strategic problem-solving and crime prevention. Given sufficient (but modest) dosages, the extra patrols reduced crime at the hot spots, but this effect was smaller in places where officers made greater use of technology.
Conclusions:
Basic applications of mobile computing may have little if any direct, measurable impact on officers’ ability to reduce crime in the field. Greater training and emphasis on strategic uses of IT for problem-solving and crime prevention, and greater attention to its behavioral effects on officers, might enhance its application for crime reduction.
The last few decades have been a period of particularly rapid technological change in policing. There have been many important developments with respect to information technologies (ITs), analytic systems, video surveillance systems, license plate readers, DNA testing, and other technologies that have had far-reaching effects on police agencies. Technology acquisition and deployment decisions are high-priority topics for police (e.g., Koper, Taylor, & Kubu, 2009), as law enforcement agencies at all levels of government are spending vast sums on technology in the hopes of improving their efficiency and effectiveness.
While these changes have much potential for improving police management and operations, it is not clear if or how much they have improved police effectiveness (e.g., Byrne & Marx, 2011; Koper et al., 2009). Research on police technology has tended to focus more on functionality and outputs—for example, whether a technology works and makes a process faster—than on effectiveness in reducing crime or improving service to citizens. And the evidence that is available on technology and police performance suggests that technology’s impacts may be limited or offset by many factors ranging from technical problems to officer resistance. Developing a better understanding of technology’s impacts and how they can be optimized is thus an important challenge for police agencies, particularly those hoping to leverage new technologies as a force multiplier to offset budget and staffing limits.
This study assesses the uses and crime control effects of mobile computing technology, which has been among the most central new technologies introduced into policing over the last few decades. As part of a hot spots patrol case study conducted in a suburban jurisdiction, we used both quantitative and qualitative methods to examine how officers used mobile computing for hot spots patrols and to assess whether their use of this technology made the patrols more effective in reducing crime. The study extends research on police technology by probing more deeply into specific ways that IT shapes officer behavior (or fails to do so) and by investigating whether its use by patrol officers has a direct effect on crime. In the process, the study also yields insights into patrol activities that may be more or less effective in reducing crime at hot spot locations. More generally, the study adds to the limited literature on the effects of IT on outcomes in the public sector and highlights some of the organizational issues that are important to understanding the relationship between IT and performance.
IT, Mobile Computing, and Police Effectiveness
ITs within police agencies include a wide array of databases and data systems (and their supporting hardware and software) for storing, managing, retrieving, sharing, and analyzing information both within and across agencies. Common IT components in police agencies include records management systems (RMS) that capture criminal incident records, computer-aided dispatch systems that record and assign calls for service, and various other databases that contain information on persons, groups, personnel, and other matters.
By improving the ability of police to collect, manage, and analyze data, IT has enhanced the administrative efficiency of police organizations, improved their apprehension capabilities, and given them the ability to more precisely and proactively target the people, places, and problems that contribute most to crime and disorder in their communities (e.g., see Groff & McEwen, 2008). Nonetheless, the relationship between IT development and police performance is not entirely straightforward. For example, in a national study of police agencies over the period of 1987–2003, Garicano and Heaton (2010) found that increases in the application of IT by police were not associated with improvements in case clearance rates or crime rates. Brown (2014) reported similar findings in a national study focusing on more recent changes (2003–2007) in police IT capabilities and clearance rates. However, Garicano and Heaton also found evidence that IT was linked to improved performance (i.e., higher clearance rates and/or lower crime rates) when complemented with other organizational changes including greater use of specialized units, higher educational and training requirements for staff, and managerial practices indicative of Compstat (a managerial approach that stresses data-driven problem identification and assessment, geographic resource allocation, problem-solving, and greater accountability for managers). Hence, as others have found in the study of technology in the private sector (e.g., see Brynjolfsson & Hitt, 2000; Milgrom & Roberts, 1990), the impact of IT (and other technologies) may often depend on other organizational changes and practices that help police leverage this technology in the most strategically optimal ways (we return to this issue below).
For line-level officers, a major advancement in IT has been the spread of mobile computers and data terminals that give officers wireless access to information in the field and allow them to file reports remotely. According to the national Law Enforcement Management and Administrative Statistics (LEMAS) survey, more than half of local police agencies reported having in-field computers or terminals for their officers as of 2007 (Burch, 2012, p. 16; Reaves, 2010, p. 23). More than 90% of local police departments serving populations of 25,000 or more had such capability at that time, as did more than 85% of sheriffs’ offices serving populations of at least 100,000. LEMAS data show that most agencies use their in-field computers and terminals for writing reports, and many also use them for other communications. Information commonly accessible to officers through these computers and terminals include motor vehicle records, warrants, calls for service, criminal histories, protection orders, interagency information, the Internet, and crime maps.
Mobile computing technology enhances officers’ capabilities in many ways. For example, it improves their access to real-time data on crime and other events; their time deployed in the community; their ability to identify persons, vehicles, and places of interest (which potentially improves both reactive and proactive field work as well as officers’ ability to identify potential safety threats); their ability to identify and locate suspects in criminal investigations; their problem-solving capabilities; and the quality of information they can provide to the public (e.g., see Groff & McEwen, 2008; Koper et al., 2009). 1 Yet, as with many police technologies, it is unclear whether mobile computing has had a discernible impact on police efforts to control crime or improve other service to citizens. (We highlight the former issue here since the study described below focuses on crime control outcomes.) A few studies suggest that advances in IT have improved case clearances by detectives (Danziger & Kraemer, 1985; Ioimo & Aronson, 2003; but also see Zaworski, 2004) and that mobile computing has increased recoveries of stolen vehicles by patrol officers (but not clearances for these crimes; Nunn, 1994). To our knowledge, however, there has been no direct study of whether officers’ use of IT in the field helps them reduce crime. This is an important issue that reflects on the value-added and cost efficiency of this technology. Moreover, investigating this issue can potentially yield insights into strategies for more effective use of IT.
Studies of police IT (and mobile computing in particular) have focused largely on how this technology affects officers’ attitudes, behaviors, and outputs. These studies show that while officers generally have positive attitudes toward IT improvements, the effects of IT have been mixed with respect to improving productivity, case clearances, proactive policing, community policing, problem-solving, and other outcomes (e.g., Agrawal, Rao, & Sanders, 2003; Brown, 2001; Brown & Brudney, 2003; Chan, Brereton, Legosz, & Doran, 2001; Colvin, 2001; Danziger & Kraemer, 1985; Groff & McEwen, 2008; Ioimo & Aronson, 2003, 2004; Nunn, 1994; Nunn & Quinet, 2002; Palys, Boyanowsky, & Dutton, 1984; Rocheleau, 1993; Zaworski, 2004). Extrapolating from this research would thus seem to raise questions about the direct crime control effects of this technology.
In general, the aforementioned studies highlight a number of factors that can offset the potential benefits of IT for field officers. For example, officers may be hampered by technical difficulties and the complexities of using mobile computing systems, particularly when the systems are new. Further, the adoption of new IT systems often leads to more extensive reporting requirements for officers. Combined, these factors may negate expected time savings, lessen (or fail to improve) time for interacting with citizens and engaging in proactive work, and create frustration and dissatisfaction for officers (e.g., Chan et al., 2001; Colvin, 2001; Ioimo & Aronson, 2004). Some also worry that an overemphasis on technical skills and computer literacy may come at the expense of skills in dealing with people in the community (e.g., Palys et al., 1984).
Moving beyond these concerns, officers may also fail to use IT in the most strategically optimal ways for reducing crime. Although research suggests that police are most effective in reducing crime when they focus their efforts on high-risk places and groups and use problem-solving strategies tailored to specific issues (e.g., Braga, Papachristos, & Hureau, 2012; Braga & Weisburd, 2012; Lum, Koper, & Telep, 2011; National Research Council, 2004; Weisburd, Telep, Hinkle, & Eck, 2010), police may not regularly employ technology toward these ends in practice. Perceptions and uses of technology are highly dependent on the norms and culture of an agency and how officers view their function (i.e., technological “frames” in the words of Orlikowski & Gash, 1994; also see Chan et al., 2001). Because officers continue to frame policing in terms of reactive response to calls for service, reactive arrest to crimes, and adherence to standard operating procedures, they use and are influenced by technology to achieve these goals (Lum, 2010). To illustrate, in a recent study of four large urban and suburban police agencies in the United States (including the agency featured in this study), surveyed patrol officers were much more likely to use IT to guide and assist them with traditional enforcement-oriented and reactive activities than for more strategic, proactive tasks (Koper, Lum, Willis, Woods, & Hibdon, 2015). Across the agencies, for example, 42–74% of patrol officers reported using IT often or very often to locate persons of interest and 63–81% in most agencies did so to check the call history of a location or person before responding to a call. In contrast, only 14–50% used IT often or very often to determine where to patrol between calls (indicative of a focus on high-risk places) or to determine how to respond to a crime problem (indicative of problem-oriented policing). Hence, police may not fully capitalize on the aspects of technology that enable them to do things that could make them more effective (Groff & McEwen, 2008). This tendency is likely reinforced by the limitations of police IT systems, in terms of available data and functionality, for facilitating problem-solving tasks (e.g., Brown, 2001; Brown & Brudney, 2003; Nunn & Quinet, 2002).
Other Lessons From Research on IT and Performance
Although our focus here is on IT and policing, similar concerns and uncertainties have arisen from efforts to assess the effects of IT on performance and productivity more broadly in the public and private sectors, where studies have yielded mixed results with regard to IT’s impacts on aggregate-level productivity, organizational performance, and client outcomes (e.g., see Brown, 2014; Brynjolfsson, 1993; Brynjolfsson & Hitt, 2000; Foley & Alfonso, 2009; Garg et al., 2005; Garicano & Heaton, 2010; Goolsbee & Guryan, 2006; Lee & Perry, 2002; Lehr & Lichtenberg, 1999; Stiroh, 2002; Triplett, 1999). Scholars have offered a number of hypotheses for this disconnect which may also be applicable to police use of IT. Brynjolfsson (1993), for example, has suggested that IT may fail to improve performance or have effects that are difficult to detect for several reasons including mismeasurement (e.g., technology may improve quality of service in less tangible ways that are not measured by standard performance metrics like crime rates), learning curves that delay technology’s impacts, mismanagement of IT implementation (e.g., see Goldfinch, 2007), and the possibility that technology may improve organizational processes without improving outcomes (e.g., improving the speed and accuracy of police reporting without clear effects on crime control; in other public sectors, see Garg et al., 2005; Goolsbee & Guryan, 2006).
Another key factor is the notion of “complementarity” (Milgrom & Roberts, 1990), which suggests that technology may not have much positive impact on its own (or may even have adverse effects) unless it is accompanied by other organizational changes—in work processes, procedures, training, management, organizational structures, and the like—that enable an organization to effectively leverage new technological capabilities. Garicano and Heaton’s (2010) research on police performance and IT, discussed above, provides one example (see Brynjolfsson & Hitt, 2000, for a review of evidence on this idea in the private sector). Yet, efforts to make such organizational changes may be less likely in public sector organizations like police agencies because, as noted by Brown (2014), public sector organizations may often adopt technology for social and political reasons (e.g., to appear more competent and legitimate) that are not linked to clear goals for enhancing performance.
Other organizational perspectives stress the role of human agency and social interpretations as reasons for varying outcomes associated with technology in organizations (e.g., Boudreau & Robey, 2005; Orlikowski & Gash, 1994). This view, which is linked to our discussion of technological frames above, stresses the interaction of technology with human agents. While technology shapes human actions, people in organizations also enact technology in different intended and unintended ways (e.g., in innovative or perfunctory ways) that can produce varying outcomes, both positive and negative. In this manner, technology and human actors both affect and transform one another in their effects on performance.
Our study integrates and applies these ideas to the analysis of police and IT. Consistent with the human agency perspective, we explore how the capabilities of mobile IT combine with officers’ views of both technology and their jobs to shape police use of IT in the field. We then use these findings to help interpret the relationship, or lack thereof, between measures of officer IT use and measures of the effectiveness of police patrol. We also consider what these findings suggest about the types of organizational changes, or complementarities, that may be needed to help police agencies optimize the use of IT among their line-level patrol staff. In the process, we also expand the limited research on IT and performance outcomes in public sector organizations more generally (see Brown, 2014).
Mobile Computing and “Hot Spots” Policing
This study examines the uses and impacts of mobile computing technology in the context of hot spots policing, a term that refers to police interventions (e.g., patrol, enforcement crackdowns, and prevention efforts) focused on small areas or very specific places where crime is concentrated. Hot spots policing is grounded in research showing that roughly half of crime occurs at 5% or less of a locality’s addresses and intersections (e.g., Sherman, Gartin, & Buerger, 1989; Weisburd, Bushway, Lum, & Yang, 2004) and that this concentration tends to be stable over time (e.g., Braga, Papachristos, & Hureau, 2010; Koper, Egge & Lum, 2015; Weisburd et al., 2004). This strategy has become an increasingly common one used by local police agencies, particularly those serving large jurisdictions (Burch, 2012, p. 16; Koper, 2014; Reaves, 2010, p. 22; Weisburd & Lum, 2005). There is also substantial consensus among researchers, based on numerous experimental and quasi-experimental evaluations, that hot spots policing is effective in reducing crime and disorder (e.g., Braga et al., 2012; Lum et al., 2011; NRC, 2004; Weisburd & Telep, 2014), though questions still remain about the most optimal dosages and strategies for police to employ at these locations.
Studying officers’ use of mobile computing in hot spots can yield insights into whether and how this technology enhances officers’ effectiveness in conducting hot spots policing. For example, mobile computing technology might facilitate problem-solving approaches, which can be particularly effective for hot spots (Braga et al., 2012; Taylor, Koper, & Woods, 2011), by providing officers with easy-to-access information in the field about very specific addresses within a hot spot block, or by giving officers the capability to search past information on crimes and calls for service to assess patterns at the location. More traditional, enforcement-oriented applications of IT (e.g., running names or license plate numbers through agency data systems) might also be more productive in hot spots, given that these locations tend to draw more activity and potentially troublesome people. Hence, even the most basic, marginal effects of IT in the field may be more discernible in this context.
We investigated these issues as part of a study that examined the effects of hot spots patrols in a suburban jurisdiction. Our specific questions about technology, which focused primarily on mobile computing, included the following: How do officers use technology at hot spots? What specific technologies do they typically use, how extensively do they use them, and which do they find to be most useful? How does technology shape officers’ actions and affect the outcomes of their efforts? Finally, how might police use technology more effectively for hot spots policing? We examined these questions using an embedded mixed methods design (Creswell, 2014) in which we complemented quantitative analyses of technology uses and impacts with qualitative data gathered from interviews and observations conducted with officers during and after the patrol intervention.
In the following sections, we first describe the design and implementation of the patrol study. We then describe officers’ use of technology at the hot spots based on interviews with the participating officers and analysis of activity logs that they filled out during the intervention. Finally, we assess the impacts of technology on hot spots policing using both qualitative analysis of the officer interviews and quantitative analyses that examine whether changes in crime at the intervention hot spots were linked to officers’ use of technology.
Study Setting, Design, and Intervention
The study was conducted with a suburban county police agency that has over 1,000 officers and serves a population of more than 1 million in the eastern part of the United States. While diverse in terms of its racial and ethnic makeup, the agency’s jurisdiction is relatively affluent (less than 10% of the population is below the poverty line) and has a relatively low crime rate (in 2012, the county’s Uniform Crime Reports index crime rate was roughly 1,400 per 100,000 persons, considerably lower than the average of 2,281 crimes per 100,000 persons for metropolitan counties generally). The county’s geography is also large and diverse, covering over 400 square miles with a mix of highly and less urbanized areas.
The study was implemented during the fall of 2012 in one of the agency’s highest crime districts as part of a broader effort to design, implement, and test a hot spots strategy that would optimize the agency’s resource allocation and effectiveness by increasing patrol presence, activity, and technology utilization at hot spot locations that accounted for a disproportionate share of crimes and calls for service. For this purpose, the agency and research team conducted an experiment in which a selected group of crime hot spots were randomly assigned to receive or not receive special patrols during a period of nearly 3 months. We describe the design, implementation, and results of the patrol experiment in order for readers to understand the full study context and analysis. As explained below, our investigation of mobile IT’s uses and impacts was built into the patrol study but not based on random assignment. 2 We gathered quantitative and qualitative data on officers’ activities in the experimental locations to better understand the frequency, nature, benefits, and limitations of officers’ uses of technology in the field. In addition, we conducted a series of nonrandomized and more exploratory analyses to determine whether the effects of the special patrols on crime varied based on how the project officers used technology in the hot spots. More specifically, we exploited natural variation in officers’ use of technology across the study locations to investigate whether locations that received a style of extra patrol that involved heavier application of technology experienced changes in crime that differed (for better or worse) from those in locations that received a style of extra patrol with less emphasis on technology use, conditional on patrol dosage, and relative to control locations that did not receive any enhanced patrols. In our discussion, we interpret these patterns with reference to our quantitative and qualitative analyses of how officers used technology in their patrol work.
The design of the study began with a street segment analysis of crime incident reports and calls for service in the study district during 2010 and 2011. Based on this analysis, the research team consulted with district commanders and officers on the selection of 18 candidate locations for the study. 3 The selected hot spots consisted primarily of apartment complexes, retail shopping centers, parking lots, and other types of business, commercial, or residential locations (e.g., locations with restaurants and bars or mixed-use locations with both residential and commercial areas). A few also included small wooded areas known for homeless encampments or homeless shelters. Given the suburban nature of the jurisdiction, incident reports and calls at these locations largely pertained to theft offenses (e.g., larceny and shoplifting) and other general forms of crime and disorder such as simple assaults, disorderly conduct, drug violations, and destruction of property. More serious forms of crime such as robbery, burglary, and sex offenses also occurred at these locations, though not frequently.
To assign the locations to the experimental (intervention) or control (no intervention) groups, the research team matched the 18 candidate hot spots into 9 pairs (or statistical “blocks”) based on counts of crime incidents and calls for service during 2011, the location type (i.e., retail/commercial or residential), and the types of crime incidents that occurred in the location during 2011 (i.e., predominantly theft or general crime and disorder). 4 From each pair, one hot spot was randomly assigned to the experimental group and the other to the control group, thus producing nine experimental locations and nine control locations. After randomization, there were no statistically significant differences between the two groups of locations with respect to size or levels of crime and disorder: Both groups averaged 65 reported incidents and 133–135 calls per location during 2011, and they averaged from 0.002 to 0.006 square miles in size.
The hot spot patrols were carried out by a group of nine officers. Six were members of a special unit that engaged in various types of proactive work (e.g., bike patrols, investigations, stakeouts, etc.) throughout the district. The officers from this unit worked in pairs, and each pair had responsibility for working two of the experimental hot spots. In addition, three district patrol officers participated in the experiment, with each assigned to cover one experimental hot spot. However, there were times during the experiment when officers worked hot spots that were assigned to others, depending on need. Project officers were initially asked to conduct three 15- to 30-min stops per day at their assigned hot spots (see Koper, 1995; Telep, Mitchell, & Weisburd, 2014), but they were given discretion to conduct fewer but longer visits to the locations if they were engaged in an activity that warranted a longer stay (e.g., problem-solving) or if that seemed most appropriate based on conditions (i.e., activity levels) at the locations.
Due to resource limitations and other needs, the patrol dosages delivered to the experimental locations were modest (the officers conducting the hot spot patrols were not dedicated exclusively to the project). 5 During the 11-week project (spanning from September through November 2012), officers made 168 visits to the target hot spots. This averaged to almost 19 visits per location over the course of the 11-week experiment, or close to 2 visits per week to each location. 6 The visits, which mostly occurred between the hours of 3 p.m. and 1 a.m., averaged 26 min. 7 We control for differences in dosage levels across the experimental hot spots (which were notable) in the outcome analysis below. However, our assessment focuses primarily on how officers used technology at the hot spots and how the effects of the patrols, controlling for dosage, varied based on these uses.
Documenting Technology Use
Participating officers used log sheets (see Appendix) developed by the research team to identify the hot spots they visited, record the times when they entered and left the locations, and document the activities they conducted. For the latter purpose, the logs had a checklist of proactive tactics grouped into the categories of extra proactive visibility (e.g., foot or bike patrol, surveillance in a prominent location), using IT (see list below), working with offenders and victims at places (e.g., conducting knock and talk visits with offenders or repeat victims or following up on incidents already reported), and proactive engagement and problem-solving (e.g., engaging with property managers or business owners or conducting traffic stops and field interviews). Officers were asked to check off each type of activity that they conducted while in the hot spot, though the activities they chose were left to their discretion. Finally, the log had space for officers to provide a brief synopsis of actions they took and to record any noteworthy results (e.g., detecting a wanted person or making an arrest).
The technology section of the checklist (which was developed in consultation with the participating officers and others from the agency) included several items for possible application of mobile computing within the hot spots. Through their mobile computers, officers had access to federal (i.e., National Crime Information Center), state, and local databases on wanted persons, criminal records, stolen vehicles, and state vehicle registration records. In addition, the agency’s RMS, implemented in 2010, enabled officers to file reports remotely from the field and provided them with wireless access to a wide range of agency data. Using this system, officers could search a name, license tag number, address, or phone number, for example, and retrieve up-to-the-minute information on all incident reports, calls for service, and field contacts associated with that entry in the agency’s records. Other resources like bulletins and crime maps were also accessible remotely through the agency’s intranet system. Finally, some officers had access to the Law Enforcement Information Exchange (LInX), a regional data sharing system that contains incident and field contact records submitted by participating agencies throughout the region. 8
Based on these capabilities, the activity log included entries for (1) conducting deeper investigation of specific individuals or addresses; (2) examining recent calls, incidents, and other events in the hot spot; (3) checking license plates of moving or parked vehicles; 9 (4) using LInX to check suspects stopped; 10 and (5) using the agency’s automated fingerprint identification system (AFIS) to check stopped suspects who did not have identification. We used these entries, supplemented by interviews and ride-alongs with the officers, to document and assess how officers used their mobile IT capabilities for the patrols.
In planning and training sessions, the research team also encouraged the officers to consider proactive and strategic ways that they might use available IT for more in-depth investigation and problem-solving. To help facilitate this, the back of the log sheet included tips, developed with the assistance of technologically savvy project officers, for using mobile computer access to the agency’s RMS and computer-aided dispatch in proactive ways (see Appendix). 11 The agency’s RMS was relatively new at the time of this project, and its implementation had been challenging (see further discussion in Koper et al., 2015). Because the agency had limited time available for training officers on the system when it was introduced, instructors had focused primarily on the mechanics of using it for entering reports. Trainers were not able to share many of the more advanced capabilities of the system with officers; in particular, it seems that there was little emphasis on its potential strategic or proactive uses for investigation and problem-solving. The tips provided on the project log sheet were thus intended to bolster officers’ knowledge of the system’s capabilities and facilitate its use for more proactive tasks.
Technology Use at Hot Spots
Analysis of Activity Logs
In general, officers were most likely to patrol the hot spots using traditional roaming patrol in their vehicles, which was an activity reported in 73% of visits. Other common activities included conducting surveillance in prominent locations (35% of visits) and conducting traffic stops (32% of visits). Officers’ most frequent use of technology in hot spots was checking the license plate numbers of moving or parked vehicles, which officers did in two thirds (66%) of their visits (see Figure 1). Officers used LInX to run checks on suspects in 20% of their visits. Further, they used the agency’s information systems to conduct deeper investigations of particular people or places during 19% of their visits, and they used these systems to examine recent calls in the hot spots during 14% of their visits. Finally, officers used AFIS to check suspects without identification during 4% of their visits. The log summaries also showed that there were two visits during which officers reported detecting wanted persons or vehicles using the agency’s information systems.
Percentage of visits in which officers used specific information technologies.
Although the officers appeared to make limited uses of technology, it is notable that they relied on technology more heavily than on many other strategies. Checking license plates was one of the most common of all activities in the hot spots, second only to vehicle patrol. Further, officers used technology to examine recent calls and conduct deeper investigation of persons or places more often than they conducted many other activities including foot or bike patrol (used in 11% of visits), discussion of problems with community stakeholders and authorities (used in 10% of visits), and various other activities that were conducted even less frequently.
Qualitative Assessments of Technology Use
To assess officers’ experiences with technology in the field, the two lead investigators conducted three focus groups with the participating officers before, during, and after the intervention. They also held additional interviews with five of the officers, primarily during ride-along observations. Supplemental qualitative data were also gathered during meetings that the investigators held with the officers and their supervisors to plan the experiment. Topics covered in the interviews and focus groups included officers’ general experiences in policing hot spots, ways in which they used technology in the field (including how technology shaped their approaches to policing the hot spots), and the benefits and limitations to using technology (e.g., in terms of efficiency and effectiveness).
The interviews and focus groups yielded a number of themes. 12 One of these highlighted the ways in which technology enhanced the officers’ efforts. Officers noted, for example, that mobile computing has made checking license plates much easier and quicker and that automated fingerprint systems have also improved their efficiencies. Officers working in two-person units found the mobile computing capabilities to be particularly helpful because one officer could run vehicles and associated people through various data systems while the other one drove and observed the location. Using their mobile computing systems, officers could collect a good deal of information about vehicles or subjects before making a stop, which might shape their approach to an encounter. In some cases, this could even reduce the necessity of making unnecessary traffic stops that might otherwise create friction with community members. Seeing information about previous police contacts with stopped subjects or previous incidents at a location could also be valuable to officers in formulating their approach to an encounter.
At the same time, there were also limits and potential drawbacks to their use of technology. As noted above, the officers detected few wanted persons or vehicles and made few apprehensions using their mobile IT. Some officers also cautioned against an overreliance on technology. As one stated, “There are three generations of officers when it comes to technology: officers who won’t use it, a mixed middle-of-the-road group, and young officers who rely too much on technology.” Elaborating further, the officer noted that officers “… who rely too much on technology lack the ability to interact well with people …. technology is depersonalizing.” Although officers agreed that good technology was important to doing their job well, one observed that “you can often learn more from a quick conversation with someone than from technology.” Similarly, officers emphasized that common sense and visual cues were particularly important in their work. Some even suggested that computer work could be distracting and pose a safety hazard, especially when an officer was unfamiliar with the area. Ultimately, as one puts it, “There needs to be a balance between the technology and human work. There can be an overuse of technology.” Consistent with this notion, some officers seemed to place more emphasis on activities like talking with juveniles, following up on particular crimes or other pieces of information, or showing visible presence.
A third theme was that officers tended to emphasize the use of technology for traditional and reactive tasks rather than strategic, preventive problem-solving. Officers’ most common approach was to enter the hot spots and respond visually to behaviors and conditions; it was less common for them to use technology beforehand to analyze problems and develop strategies for the locations. Hence, their most common uses of IT were to check license plates or run checks on people they had stopped. Further, when asked what technologies would be particularly important to them, officers emphasized technologies that would help identify people (e.g., AFIS) or allow them to “get deeper into an investigation.” When officers did use IT to do research on a particular place, they focused on locations given to them by other people (through referrals, tips, or bulletins) or places they had identified through social media in connection with a specific incident (e.g., a place where a potentially troublesome party was going to occur). Although some officers saw the potential of using mobile IT to better understand problems at their assigned locations (one officer, e.g., used the mobile IT system to examine call histories and past crime reports for specific apartments at an assigned apartment complex), this did not lead to systematic, in-depth problem-solving efforts.
A fourth theme is that there were both technological and organizational limits to using IT in more strategic ways in the field. In terms of technical limitations, several officers noted that they did not receive highly customized information—and did not know how to access such information—from the RMS. For example, a “top 10” offender or address list within a hot spot could not be obtained, although the call history for a particular address could. And despite the technology tips developed for the project, officers found it difficult and cumbersome to search for many things in their RMS. 13
When asked what might encourage more strategic and in-depth use of technology throughout the agency, the project officers mentioned more training, knowing specific “tricks” to operating the system (especially for obtaining more in-depth information on people, places, or incidents), stressing the benefits of the system to officers (e.g., its ability to provide officers with real-time information on subjects they encounter), and using e-mail to disseminate information about these matters. Officers also pointed to motivated supervisors, knowledgeable field training officers, and personal motivation as key determinants of whether officers use technology proactively toward strategic goals (though they also felt that some officers, particularly older ones, would still be unlikely to adapt in this manner).
Overall, it appeared that mobile IT enhanced the work of officers who participated in the experiment, although it was more often used in support of traditional proactive work such as making traffic stops, conducting field interviews, and supporting investigations. Officers tended to adapt to the technology based on their own training and knowledge, and many tried to carry out proactive policing without technology in crime hot spots. However, the officers rarely used technology to strategically or systematically direct their work outside of these operational modes, nor did the IT system itself motivate officers to be proactive. Rather, officers characterized the system more as a resource to help them do more proactive work. The majority of their efforts (notably stops) were driven by visual cues and observations rather than computer information.
Impacts of Patrol and Technology on Crime
After examining how officers used technology and assessing those uses qualitatively, we estimated a series of statistical models to assess the quantitative effects of the patrols on crime and disorder at the experimental locations and to determine if and how those effects varied based on officers’ use of technology. To do this, we used total crime incident reports as our primary outcome measure and modeled weekly incident counts in the experimental and control locations using a panel database in which each observation corresponded to a given hot spot (i) during a given week (t). 14 This produced a panel database of 18 hot spots × 11 project weeks = 198 observations. 15 We modeled crime reports for each hot spot per week as a function of: a patrol intervention indicator showing whether the hot spot was one of the experimental locations receiving enhanced patrol (we used multiple patrol indicators as described below); the blocking indicator used for random assignment, which captures the combined effects of the location type and its historical crime patterns (the block indicators also serve as fixed effects for pairs of matched locations); a seasonally lagged Time 1 measure of crime at the hot spot during the same week of the prior year (i.e., 52 weeks lagged from week t); 16 and other measures, including technology indicators, as discussed below. Based on preliminary tests of Poisson and negative binomial count distributions for the outcome measure (e.g., see Cameron & Triverdi, 1986), we used models of the latter, which provided a better fit to the data as assessed by a likelihood ratio test. The models were estimated with generalized estimating equations (Zeger, Liang, & Albert, 1988; also see Allison, 2005) that allowed for dependence between observations from the same hot spot. 17
Descriptive Measures
Table 1 presents key implementation measures for the experimental locations. Visits per week to the experimental hot spots averaged about 1.7 and ranged from 0 to 8. Total minutes spent by officers at these locations averaged 42 per week within a range of 0–225. Our primary measure of technology use is IT uses per visit as measured by the officer logs. This measure reflects how extensively officers used technology at the hot spots conditional on how frequently they visited (i.e., a measure of technology use relative to patrol dosage). It was calculated as the number of total technology uses for the week divided by the number of visits for the week (conditional on the number of visits being greater than zero). 18 This measure averaged 1.4 uses per visit and ranged from 0 to 4 (corresponding to the four types of technology use on the officer log).
Key Implementation Measures for Experimental Locations.
Note. n = 99 hot spot-week observations for the measures of visits and minutes per week. The technology uses per visit measure is based on 68 hot spot weeks with one or more visits.
Crime measures for the experimental and control locations are displayed in Table 2. For each group, the table shows the average weekly number of crime reports during the project period and the average weekly number that occurred during the corresponding weeks of the prior year (i.e., the seasonally lagged Time 1 crime measure). Hot spots in both groups averaged about two crimes per week during both the intervention and seasonally lagged periods. During the project period, the number of incident reports per week ranged from zero to six in the experimental locations and from zero to nine in the control locations.
Key Crime Measures for All Locations.
Note. n = 99 hot spot-week observations for the experimental group and the control group (for a total N of 198).
Model Results
Two basic models of the patrols’ effects, without consideration of technology, are displayed in Table 3. Model 1 shows the overall change in the experimental locations relative to the control locations (i.e., treatment as assigned) during the project period. Overall, crime reports declined 11% in the experimental hot spots, though this decline was not statistically significant (p > .10). 19 Because the delivery of the intervention varied considerably across sites, we also estimated a second version of the model in which the experimental locations were divided into high- and low-dosage groups. The high-dosage locations were designated as those that were above the median (i.e., the top four treatment hot spots) on the average number of minutes of special patrol per week. By this criterion, the high-dosage locations averaged 47–90 min of patrol per week during the experiment, and the low-dosage locations averaged 8–45 min per week. As shown by Model 2, the high-dosage locations experienced a statistically significant 24% reduction in crime reports relative to the control locations, while the low-dosage locations showed no change. This suggests that the patrols reduced crime when the dosage was sufficiently high, though this finding should be interpreted more cautiously since the exact patrol dosages were not randomly assigned. 20
Impacts of the Hot Spot Patrols on Crime Incident Reports.
Note. Models include statistical block effects and an intercept term, which are not shown. Estimates are based on generalized estimating equations that control for first-order autoregressive dependence between observations from the same hot spot. p Levels are based on Z-score statistics. N = 198.
We conducted two sets of nonrandomized analyses to investigate whether the effects of the patrols varied based on officers’ use of technology, conditional on patrol dosage. For the first set, we divided the experimental locations into high and low patrol dosage groups based on the criterion described above. We also rank ordered the experimental locations based on their average level of technology use per visit during the intervention period and divided them into high- and low-technology groups. Cross-referencing the patrol and technology rankings produced four subgroups among the experimental locations: places that received high patrol dosages with high levels of technology use by officers (two locations), places that received high patrol dosages with low levels of technology use by officers (two locations), places that received low patrol dosages with high levels of technology use by officers (three locations), and places that received low patrol dosages with low levels of technology use by officers (two locations). 21 We then estimated models of the patrol effects in each of these subgroups relative to their matched and randomly assigned control locations.
The results of these models are presented in Tables 4 and 5. As a caveat, note that the sample sizes are small in these analyses (44–66 location-by-week observations per model); accordingly, we emphasize the direction and magnitude of the coefficients as much as their statistical significance levels. Model 1 in Table 4 shows that the high dosage–low technology locations experienced a 45% reduction in crime reports relative to their matched and randomized control locations (this difference had a marginal statistical significance level of p = .08). Model 2 of Table 4 shows that the patrols also reduced crime in the high dosage–high technology locations, but this effect (though statistically significant) was considerably smaller at 14%. Similarly, the models in Table 5 for the low-dosage locations show that the patrol effects, though not statistically significant in either model, appeared more beneficial in the locations where officers made less use of technology (these locations experienced a slight reduction in crime relative to their control locations). Hence, models for both the high- and low-dosage locations suggest that greater use of IT did not enhance the effects of the patrols; that is, locations that received a style of extra patrol that involved heavier use of technology did not experience greater crime reductions than those that received a style of extra patrol that involved less use of technology, conditional on patrol dosage, and relative to their respective control locations.
Impacts of the Patrols on Crime Incident Reports at High-Dosage Locations by Level of Technology Use (Measured by Technology Uses per Visit).
Note. Models include statistical block effects and an intercept term, which are not shown. Estimates are based on generalized estimating equations that control for first-order autoregressive dependence between observations from the same hot spot. p Levels are based on Z-score statistics. n = 44 for both models.
Impacts of the Patrols on Crime Incident Reports at Low-Dosage Locations by Level of Technology Use (Measured by Technology Uses per Visit).
Note. Models include statistical block effects and an intercept term, which are not shown. Estimates are based on generalized estimating equations that control for first-order autoregressive dependence between observations from the same hot spot. p Levels are based on Z-score statistics. n = 44 for Model 1 and n = 66 for Model 2.
The preceding analyses have the benefit of comparing each subgroup of treatment locations to a set of matched control locations that were randomly assigned to not receive the additional patrols. However, a limitation to these analyses is that the level of IT use was not a randomly assigned condition, and there was a tendency for officers to make more use of technology in places that had higher levels of crime (likely because there were higher levels of pedestrian and vehicle traffic and activity at these locations, thus resulting in more opportunities for the officers to use IT to run checks on people and vehicles). Consequently, the high- and low-technology subgroups may not have been highly comparable locations even within each dosage level.
Therefore, we estimated an additional model for which we pooled data across all locations and included additional variables for levels of crime and officer activity at the hot spots. In addition to the statistical blocking variables and the measure of seasonally lagged crime, the model includes measures of average weekly crimes at each hot spot during the 11 weeks preceding the intervention and the average number of nontechnology activities that officers conducted per visit at each intervention hot spot during the program period (e.g., traffic stops, field interviews, and talks with business owners and property managers). The latter variable provides some additional control for the possibility that officers were more active (with technology and nontechnology activities) in locations that had higher levels of crime and citizen activity, while also controlling for the potential impacts of officers’ nontechnology activities. 22
For this model, we used the average weekly dosage (in minutes) of extra patrol in the hot spot as the patrol measure and the technology use per visit indicator, averaged over the intervention weeks, as the main technology measure. In addition, we included a patrol and technology use interaction term to test whether the effects of the patrols varied with officers’ level of technology use.
As shown in Table 6, higher dosages of patrol reduced crime at the hot spots. The main effect of technology use, in contrast, was negative but not statistically significant. More importantly, the patrol–technology interaction term was positive and statistically significant. Consistent with the subgroup models above, this suggests that the effects of the patrols were not enhanced and seemed less beneficial in places where officers made greater use of technology.
Combined Model of the Impacts of Hot Spots Patrol, Technology Use, and Their Interaction on Crime Incident Reports.
Note. Models include statistical block effects and an intercept term, which are not shown. Estimates are based on generalized estimating equations that control for first-order autoregressive dependence between observations from the same hot spot. p Levels are based on Z-score statistics. N = 198.
Discussion and Conclusions
In summary, this study extends research on police technology by further illuminating specific ways that patrol officers use mobile computing in the field and by testing, albeit in a tentative way, whether their use of this technology helps them directly reduce crime. Investigating these issues in the context of hot spots policing yields insight into the actual and potential value of mobile computing for enhancing a successful patrol practice, and it also provides an opportunity to assess this technology’s crime control effects in way that is less likely to be confounded by variations in patrol practices.
The officers in this study used mobile IT primarily for checking automobile license plates and running checks on people they encountered in the course of activities like traffic stops and field interviews. Technology was an important tool that officers used to support proactive policing, but they tended to use it in more traditional ways that emphasized surveillance and enforcement. In contrast, officers did not often use technology in ways that emphasized problem-solving and crime prevention.
Further, our examination of the effects of the patrols indicated that greater use of technology was associated with weaker rather than stronger crime prevention effects. In other words, greater use of technology did not make officers more effective in reducing crime; if anything, the results suggested that officers were less effective when they used technology more extensively. These results can arguably be interpreted in multiple ways. One possibility is that officers’ use of technology was largely in reaction to the level of activity in the locations. Hence, officers may have tended to use technology more extensively, particularly for running checks on vehicles and people, in places with higher levels of vehicular and pedestrian traffic, and the patrols may have tended to have less impact in these locations. Under this interpretation, technology use did not appreciably enhance or undermine officer performance.
Another interpretation, however, is that the results could reflect officers’ use of proactive approaches that were effective but did not rely on technology. Some officers, for example, discussed carrying out nontechnology activities like speaking with youth and discussing problems with place managers. Some also remarked that technology can be distracting when driving alone and that officers can get too preoccupied with technology at the expense of other effective activities. Further, technology may shift street-level activity to one type of activity—checking license plates. As officers indicated, this might reduce the amount of interaction between officers and people and reduce the visibility and activity of officers more generally. In these regards, it is possible that an overreliance on technology makes officers less effective, as perhaps suggested by the quantitative results.
An additional point to emphasize is that officers tended to use technology to facilitate more traditional types of proactive activities (i.e., surveillance and enforcement) rather than preventive problem-solving, which tends to be more effective in reducing crime at hot spots (Braga et al., 2012; Taylor et al., 2011). The marginal effects from these applications of technology may be limited, as evidenced by the outcome analysis and the fact that the hot spots officers made few apprehensions based on information pulled from their IT systems. Hence, while basic application of IT might have marginal effects in improving police efficiency, detection capabilities in the field, and officer safety in responding to calls, these improvements may not alone be enough to measurably enhance patrol performance in reducing crime. 23 Perhaps greater training and emphasis on the uses of IT for problem-solving at hot spots—and better design of IT systems to enable such work (e.g., Brown, 2001; Brown & Brudney, 2003; Davis, 1989; Groff & McEwen, 2008; Nunn & Quinet, 2002)—would lead to more effective use of this technology with clearer impacts. Similarly, others have noted in business contexts that IT may initially have only small effects on performance that grow over time, as businesses work through technology learning curves and develop complementary innovations that help them leverage technology more effectively (Brynjolfsson & Hitt, 2000).
In this regard, our study may also help to further illuminate complementarities that will be necessary for police to optimize the use of technology. While Garicano and Heaton’s (2010) research has illustrated some of the managerial and structural organizational changes that enhance the impact of IT in police agencies, our study may point the way toward other adjustments that are needed in terms of training, technology design, supervision, and work processes in order to enhance the use of technology at the line level in police organizations. These lessons could also have relevance to other public sectors organizations in which IT has not clearly improved outcomes (e.g., Garg et al., 2005; Goolsbee & Guryan, 2006; also see review in Brown, 2014).
As a caveat, we note that our experiment represents a case study based on small numbers of hot spots and officers working in one suburban jurisdiction during a short span of time. Accordingly, we must be cautious in generalizing our results to other places and officers. Results might differ, for instance, in places with higher levels of crime or in agencies with more advanced IT systems, more experience using those systems, and a stronger managerial commitment to the use of IT for crime reduction (Brown, 2014). 24 Also, our quantitative assessment of the impact of technology on crime at the hot spots was not based on random assignment but rather on the interaction of technology use with a randomly assigned patrol intervention, supplemented by quantitative and qualitative analyses of officer activities and technology use. Our quantitative outcome measures may have also failed to capture some ways in which IT improved the quality of service that officers provided to citizens (e.g., see Brynjolfsson, 1993), though our qualitative results yielded insights into some of the less tangible positives and negatives of IT in the field. In these regards, the study should be viewed as exploratory and as more suggestive than definitive.
Notwithstanding, the study provides insights into police use of IT at the street level and complements other studies that have illustrated the complexities and contradictions involved in assessing technology’s effects on police behavior and effectiveness. It also reinforces earlier research, suggesting that technology’s benefits in policing can depend heavily on how it is viewed, implemented, and used—insights that are consistent as well with those found in other organizational research showing how technology-related outcomes are shaped by the interaction of technology with human agents. Although technological advancements have much potential for improving policing, they may not bring about easy and substantial improvements in performance without significant planning and effort, and without infrastructure and norms that will help agencies maximize the benefits of technology. Strategizing about technology application is thus essential for police and should involve careful consideration of the specific ways in which new and existing technologies can be deployed and used at all levels of the organization to meet goals for improving efficiency, effectiveness, and agency management (Koper et al., 2015; also see Groff & McEwen, 2008). On these points, the findings and implications of this study also complement and echo those from other studies of IT in the public and private sectors.
For evaluators, future studies in this area should carefully assess the theories behind technology adoption (i.e., how and why a particular technology is expected to improve police effectiveness), the commitment of agency management to using technology to enhance performance, the ways in which technology is used throughout police agencies, and the variety of proximate and distal effects (intended and unintended, tangible and intangible) that technology may have on officer behavior, organizational processes, and community outcomes. Additionally, further research is needed to clarify what organizational strategies and complementarities—with respect to product design, training, implementation, work processes, management, and evaluation—are most effective for achieving desired outcomes with police technology.
Finally, although not the primary focus of this discussion, the study also extends research on hot spots policing. It represents one of the first studies of hot spots policing outside of a large city (an important need according to police scholars—see Weisburd & Telep, 2014), and it demonstrates the applicability of this strategy in a suburban setting (for other studies in small cities and one suburb, see Hegarty, Williams, Stanton, & Chernoff, 2014; Lum, Hibdon, Cave, Koper, & Merola, 2011; Weisburd, Hinkle, Famega, & Ready, 2011). In particular, the study shows that there is substantial concentration of crime at street segments in suburban areas (see note four), and it suggests that police can reduce crime at suburban hot spots with modest dosages of patrol (roughly three quarters of an hour to an hour-and-a-half per week on average over several weeks), though conclusions about the latter point should be tempered based on the small number of locations and short time period of the study. Our assessment of IT uses and impacts may also help police and researchers, as they strive to better understand what patrol styles and activities are most effective in reducing crime at high-risk locations in various types of jurisdictions.
Footnotes
Appendix
Acknowledgments
The authors thank the officers and commanders who took part in this study for their cooperation (we have left the identities of the participating agency and officers anonymous). We also thank Julie Grieco and Stephen Happeny for providing research assistance.
Authors’ Note
The points of view expressed in this article are those of the authors and should not be attributed to any of the aforementioned organizations or individuals.
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 research was supported by grant 2010-MU-MU-0019 from the National Institute of Justice (U.S. Department of Justice).