Police Quarterly (2022)
Abstract
Prior research has examined how social media has been integrated into law enforcement operations; however, this research has not fully explored the potential for this technology to function as an effective community policing strategy. Through the creation of a uniquely large dataset constructed with individual “tweets,” the current study analyzed Twitter use by the NYPD in 2016. Using a mixed-methods approach, this research identified linkages between social media and community policing, the level of adherence to community policing objectives, the potential for heterogeneity in usage across different NYPD patrol boroughs, and the degree of public engagement. Our findings suggest Twitter is inimitably positioned to contribute to each aspect of community policing, although its effectiveness varied across dimension typology. Second, heterogeneity was also observed across patrol boroughs, indicating both the objectives and application of social media varies within the agency. Lastly, engagement metrics reveal a few notable trends concerning police-community relations.
Citation
Thomas, A., Hatten, D., and Connealy, N. (2022). Does Police Use of Twitter Align With and Enhance Community Policing Objectives? An Analysis of the New York City Police Department’s Twitter Activity. Police Quarterly https://doi.org/10.1177/10986111211043875
Introduction
Community policing initiatives have often yielded mixed results as researchers have classified it as “too amorphous a concept to submit to empirical evaluation” (Mastrofski, 2006, p. 45). Community policing has also been viewed as a less effective reform agenda because it requires two substantial structural and programmatic changes (Mastrofski, 2006), which directly oppose the paramilitary organizational structure commonly associated with police agencies (i.e., decentralization and broadening police functions). Despite the difficulties, advocates of community policing remain resolute of its potential to improve police-community relations and reduce crime (Mastrofski, 2006). Thus, the field continues to encourage practitioners, researchers, and other stakeholders to develop an operational model of community policing that simultaneously produces the proposed benefits and stands up to empirical testing.
One of the promises of community policing is that it will have a positive impact on civic engagement by “involving the public in some way in efforts to enhance community safety” (Skogan, 2006, p. 29). The goal of increased civic engagement is to create a relationship that allows the police and members of the community to become “co-producers” of safety (Skogan, 2006, p. 29). Developing civic engagement between police and the community is more than increasing “the amount of informal “face time” between police and residents” (Skogan, 2006, p. 29) though; it also requires a significant amount of decentralization to occur. This philosophical change often makes community policing difficult to faithfully implement across law enforcement agencies. To that end, community policing is comprised of several strategies including partnerships, problem-solving, and structure, but it is also a philosophy that requires organizational transformation and systemized cohesiveness across involved entities.
Recently, researchers have begun to explore how the use of social media “is changing the landscape of public agencies and bureaucracies around the world” (Criado, Sandoval-Almazan, & Gil-Garcia, 2013, p. 319). In fact, the President’s 21st century policing task force (2015) proposed the use of social media as a tool to engage communities. Contemporary research suggests that social media outlets have now facilitated a way for a true, two-way communication model between government agencies and the community to exist (Brainard & Edlins, 2015; Heverin & Zach, 2010; Meijer & Thaens, 2013; Meijer & Torenvlied, 2016). Notably, social media technology allows the police to bypass traditional “media outlets and their gatekeeping proclivities, … offer[ing] unprecedented opportunities to control information flows, connect with public audiences, and directly, instantaneously, and continuously manage their visibility” (Walsh, 2020, p. 1139). In this capacity, social media may provide a new means to better enable the desired aims and objectives of community policing, which includes fostering two-way communication, safety co-production, and joint problem-solving efforts in real-time (see Williams et al., 2018).
Although there is a large body of research devoted to evaluating social media use by law enforcement agencies, there are only a few studies that have examined this relationship through the lens of community-oriented policing principles (see Brainard & Edlins, 2015; Liberman, Koetzle, & Sakiyama, 2013; Mayes, 2020). With only a few studies in this area, a significant gap in the research remains. Specifically, the existing research has yet to be fully grounded in a comprehensive theory that solely focuses on examining social media use via clearly defined community policing principles and themes. Additionally, research has not evaluated if social media usage aligns with community policing, and if so, in what way(s), and if the use of social media is uniform across the agency. Lastly, research has also not yet examined the reach, receptivity, and two-way communication efficacy of social media platforms, which can now be ascertained through built in metrics (e.g., favorites, retweets, etc.). Exploring the nature of these conclusions would positively contribute to our understanding of police use of social media, particularly in a society growing increasingly dependent on the facilitation of online communication. An evaluation of Twitter’s effectiveness in engaging in community policing through social media is timely, as our modern world relies on social media for real-time updates, real-time responses, and a real-time capacity to access hard to reach populations through a virtual context. Taking this into account, the promise of social media in bridging such gaps warrants review.
To contribute to the literature, we conduct a series of descriptive and regression analyses using a novel dataset. The dataset is comprised of a sample of tweets from official NYPD Twitter accounts that were coded by the authors based on their adherence to several identified community policing principles and themes. The study examines whether NYPD Twitter usage functions in a way that aligns with the principles of community policing across three objectives. First, by determining the frequency by which NYPD tweets adhere to different components of community policing. Second, by testing for potential heterogeneity across different NYPD patrol boroughs as a function of decentralization. And third, by examining the potential reach and two-way engagement of Twitter through built in metrics. The results will highlight if social media can be leveraged to meet the principles of community policing, which aspects of community policing social media is positioned to effectively meet, whether intra-departmental differences exist in the application and usage of social media, and the scope and reach of departmental tweets in connecting with a virtual audience.
Literature Review
Community Policing
The definition of community policing depends upon whom you ask. Mastrofski (2006) explains that community-policing initiatives are diverse and multifaceted, which is mostly attributed to the fact that the definition of community policing is both ambiguous and flexible. In support of this statement, Cordner (1995) explains, “community policing remains many things to many people” (p. 402). However, most researchers agree that community policing can be generally defined as any program that aims to join the police and community together in a working partnership that focuses on accomplishing a common goal (Cordner, 1995; Mastrofski, 2006; Uchida, 2005). As a result, community policing is widely considered a philosophy, as it is more aptly described as “a process rather than a product” (Skogan, 2006, p. 28).
During the 1990s, the push to incorporate community-oriented strategies surged across the US as local police agencies were able to secure supplemental federal funding to enact community-based programming (Uchida, 2005; Zhao, Scheider, & Thurman, 2002). The US Department of Justice and the Office of Community Oriented Policing Services (COPS) were responsible for awarding more than seven billion dollars to law enforcement agencies from 1995 through 2002 (Zhao et al., 2002). In addition to hiring grants, which included expanding the number of both uniformed and civilian personnel, the COPS Office also delegated the resources to fund innovative grant programs focused on technology (i.e., via the Making Officer Redeployment Effective (MORE) program) (Zhao et al., 2002). In these earlier iterations, though, community policing initiatives were generally designed to prioritize and serve the law enforcement agency by primarily increasing their capacities. Now, however, police departments are re-conceptualizing their community policing initiatives to involve two-way models that simultaneously build-up both the agency and the community.
The shift towards newer, more community-centric objectives has been driven, at least in part, by recent events highlighted by the media; these include police use of force incidents, racial profiling and discrimination, and the controversial police killings of citizens. All of which have served to divide the police and the communities they serve (see Mayes, 2020). Community policing has since become a required prerequisite and “buzzword” to any police sanctioned intervention or programming effort. Law enforcement agencies have fully shifted the community policing philosophy to prioritize two-way programming, such that every endeavor now seemingly both involves and requires law enforcement and community input, implementation, and positive reception. Community satisfaction is now just as important a criteria of intervention effectiveness as crime reduction, sometimes blurring the responsibilities and role of police. Current community policing programming now tends to focus on: (1) mending the existing rifts found within the tenuous social fabric that is responsible for holding the police and community together, (2) increasing the transparency and legitimacy of the police agency, (3) reversing the effects modernization has had on police-community relationships via the use of cars for patrol purposes, and (4) changing the communication processes between the department and community (Criado et al., 2013; Meijer & Torenvlied, 2016; NYPD, 2016; 2017b; Uchida, 2005).
Even though community policing has been practiced within the US for decades, the fact that “there is no single definition of community policing nor any universal set of program elements” (Cordner, 1995, p. 401) has made it extremely difficult to empirically assess. Most research evaluating community policing initiatives has yielded mixed results (see Gill et al., 2014; Mastrofski, 2006; Skogan, 2006). For example, in a recent systematic review and meta-analysis, Gill and colleagues (2014) examined how implementing a community-oriented policing strategy within a police department could affect the following: crime, disorder, police legitimacy, citizen satisfaction, and the fear of crime. They found that even though community policing approaches had a positive influence on three of the four aspects related to citizens’ perceptions (i.e., legitimacy, citizen satisfaction, and perceptions of disorder), it did not have the same effect on fear of crime or on actual reported crime. These findings are also congruent with Telep and Weisburd’s (2016) later assessment, which found that community policing initiatives led to an increase in both citizens’ perceptions of police legitimacy and satisfaction. This concurrence is not surprising as their work evaluated a multitude of policing strategies by assessing pre-existing systematic reviews, which examined a variety of different policing initiatives including community policing (Telep & Weisburd, 2016).
Yet, despite the inconclusive findings, American police departments are still highly encouraged to adopt community policing. A recent Gallup poll revealed that 97% of Americans believe that the police should have a good relationship with the community they serve (Crabtree, 2020). In addition, roughly 80% of Americans support the requirement that their local agency engage in techniques associated with community policing (Crabtree, 2020). The process over product approach has seemingly created a community policing umbrella that now includes all police sponsored programming, and further serves to ensure that the terminology and conceptual framework of community policing stays in vogue despite potential non-effectiveness and a potential inability to be successfully implemented in communities with the greatest need (Rukus, Warner, & Zhang, 2018). It is this type of public support that further propels community policing initiatives to the forefront of the reform agenda (see President’s Task Force on 21st Century Policing, 2015) despite mixed results regarding implementation and efficacy.
Dimensions of Community Policing as Theory
The complicated nature of community policing has plagued the evaluative efforts of previous research. Although the current research base on community policing has largely deviated from theory, early research (Bayley, 1994; Bratton, 1996; Cordner, 1995; Skolnick & Bayley, 1988) established a potential theoretical foundation by outlining a few key dimensions necessary to effectively accomplish a community agenda (see Maguire & Mastrofski, 2000). These studies defined community policing via three- or four-dimensional models, some of which shared overlapping concepts. Still, most approaches failed to provide “a clear theory of change that specifies the mechanisms by which community collaboration, problem solving, and organizational transformation should be operationalized to influence outcomes” (Gill et al., 2014, p. 420). As such, the authors further proposed that future evaluations may be better served by attempting to examine community policing with a defined theoretical framework.
Cordner’s (1995) four-dimensional model provides an exceptionally comprehensive outline of what is necessary to fulfill the goals of community-oriented policing. The four core dimensions outlined in Cordner’s (1995) work are: (1) the philosophical, (2) the strategic, (3) the tactical, and (4) the organizational. Furthermore, Cordner (1995) also discusses the common characteristics attributed to each dimension. For instance, the philosophical dimension includes carefully considering citizen input, broadening police functions, and delivering police assistance via the “service style” of policing (Cordner, 1995). The strategic dimension includes re-orienting police operations, having a geographical focus, and emphasizing prevention (Cordner, 1995). The tactical dimension includes increasing positive interactions, increasing partnerships, and including the community in problem solving (Cordner, 1995). Lastly, the organizational dimension includes re-engineering the structure of the department, altering the management principles within the department, and changing how information is shared and distributed within and by the department (Cordner, 1995). However, in order to fully adhere to each of these dimensions laid out by Cordner (1995), a complete overhaul of how police agencies interact with their constituencies must occur. Social media has been considered an effective pathway for achieving such a goal due to its ability to quickly disseminate information, connect communities and law enforcement, provide informal engagement opportunities and access to hard-to-reach populations and wider audiences, and relay departmental goals, objectives, and identity.
Social Media/Twitter
Social media is defined as “a web-based technology [that] allow[s] content creation by anyone belonging to the site and enable social interaction in the forms of networking, information exchange, collaboration, and/or deliberation” (Brainard & Edlins, 2015, p. 730). To date, some of the most popular social media outlets include Facebook, Twitter, and YouTube (IACP, 2014). Created in 2006, Twitter allows individuals to microblog, or “send brief updates via the web, mobile devices, and other applications” (Heverin & Zach, 2010, p. 1). It has become so popular that by the end of 2016 Twitter had more than 300 million users (Sparks, 2017) and currently vaunts “100 million daily active users and 500 million daily tweets” (Forsey, 2019). Twitter is different from other social media outlets in that its orientation is focused ‘in the now’ (Heverin & Zach, 2010), yet it still has the ability to form interactive relationships through conversations held in a completely online space (O’Connor, 2017).
Despite the rapid rise and adoption of social media related technology in other professional spheres, most police agencies in the US did not invest heavily in any social media platform until around 2010 (IACP, 2014). Even the US’s largest police force, the New York City Police Department (NYPD), did not fully adopt the use of social media as a department-wide initiative until 2014 (NYPD, 2016). This slower onset of adoption was contrary to the widely heralded belief that social media can encourage true two-way communication between the police and community (NYPD, 2016).
According to Mergel (2012), the dominant social media strategy any department or agency employs will differ depending upon the goals of the organization. Historically, the NYPD has utilized a “push” strategy, which involves adherence to a strict, bureaucratic model of social media application (Heverin & Zach, 2010; Meijer & Thaens, 2013; Meijer & Torenvlied, 2016; Mergel, 2010). This strategy appears to dominate social media usage amongst US police agencies, as many have reported that their primary purpose for using social media is predicated on requesting the public’s assistance with ongoing investigations (82.3%) (IACP, 2014). Beyond community engagement, the rise and growth of social media has also expanded into police practice as a potential means for surveillance and intelligence gathering (see Mateescu et al., 2015). Thus, if police are going to continue to integrate and leverage the use of social media as an agency tool, evaluation is paramount as the use, image construction, goals, and transparency of social media usage influence other spheres of law enforcement.
Current Study
The aim of this study is to examine social media (i.e., Twitter) use in law enforcement by attempting to determine if it aligns with defined community policing principles, if there is intra-departmental uniformity or heterogeneity, and the level of engagement and receptivity with the community. As such, there are three main areas of inquiry in the present study. The first research question is theoretical in nature and examines whether the NYPD’s use of Twitter is in accordance with the principles and themes of community-oriented policing as per Cordner (1995). The second research question is more evaluative and operational and tests whether the community policing principles observed are heterogeneous across patrol boroughs.1 Lastly, through an exploration of built-in Twitter metrics (i.e., retweets and favorites), we attempt to gain a sense of public engagement and reach. The results of this study will address these aspects: (a) is social media an effective means of community policing; (b) which community policing tenets does social media fulfil; (c) is this strategy used consistently and uniformly across the agency; and (d) what is the average level of engagement and reach on NYPD tweets? In sum, this study is positioned to provide important insights surrounding a growing police tool used for community relations.
Methodology
Data Sources and Sample
For this study, all individual tweets posted by an official NYPD precinct and police service area (PSA) account2 from January 1, 2016 - December 31, 2016, were collected from Twitter.3 Access to this type of data was obtained through Twitter’s application programming interface (API) (http://dev.Twitter.com). Each tweet included the following: the text and any images or other media included or linked, the precinct that sent the tweet, the date and time the tweet was sent, and other relevant meta-data. The full dataset contained a total of 42,511 tweets and individual tweets were operationalized as the unit of analysis. To achieve representation for all subsequent analyses, we took a stratified sample of the full dataset and created a smaller sample consisting of approximately 40% (N = 16,913) of the original tweets (Maxfield & Babbie, 2015). This sampling method was used to ensure an “appropriate number of elements from homogenous subsets of that population” (Maxfield & Babbie, 2015, p. 216) were included.4 This subset threshold was arrived at to allow for a conceivable amount of researcher coding, while also ensuring the sample was representative, maintained sufficient power, and included enough tweets to strengthen the external validity and minimize the potential for systematic biases (Maxfield & Babbie, 2015).
To assess NYPD Twitter use, the research design involved a mixed-methods approach. Tweets were first collected and coded qualitatively via content analysis, which was done so that the number of unique community policing principles in each tweet could be recorded.5 Then, category frequencies and interactions for the sample of tweets were analyzed using quantitative techniques. The approach interweaves the qualitative assessment of tweets to identify categories and themes, and the quantitative analysis of frequencies and relationships across the operationalized constructs.
Community Policing Principles
There were eight community policing principles identified in the literature to code for; these included: (1) enforcement, (2) crime prevention tips and safety plans, (3) community interaction (i.e., non-enforcement related), (4) dissemination of timely information, (5) causes, police/community events, and awareness, (6) blue family (e.g., faces behind the badge), (7) other units (this included both specialty units and civilian members of the service), and (8) geo-locational.6 Each tweet was classified as belonging to a specific categorical principle after a thorough read-through was conducted.7 This involved examining the text, any URLs and/or photos included, as well as considering the hashtags that were used. For example, this was one of the tweets that was coded:
A weekend of service from our #NYPDExplorers to honor the legacy of Dr. Martin Luther King Jr. Serving seniors. https://t.co/O2itGXbcQd
This tweet was classified as containing content from three different principles: (1) community interaction; (2) other units; and (3) causes, events, and awareness.8
Community Policing Thematic Dimensions
To further determine if the NYPD used Twitter as a mechanism to enhance adherence to the core principles of community-oriented policing, this research also operationalized Cordner’s (1995) four-dimensional model of community policing. Thus, each of the eight operational categories were assigned as belonging to one of the four thematic dimensions outlined in Cordner’s (1995) work. Here, the philosophical dimension (i.e., the aspect that focuses on incorporating citizen input, broadening the scope of police functions, and tailoring police services to the needs of a particular community) included tweets that contained blue family content and content relating to causes, police/community events, and awareness. The strategic dimension (i.e., the element that includes changing the organizational focus from responding to events via a reactive mentality to one that addresses potential issues via a preventative and proactive orientation; this dimension also proposes utilizing alternative patrol approaches and assigning responsibility based on geographic orientation) was identified as including crime prevention tips and safety plans, as well as all geo-locational tweets.
The tactical dimension (i.e., the component that transforms specific polices into practices via positive interactions and partnerships that focus on collective problem solving) was comprised of tweets that contained aspects of community interaction and those that focused on enforcement. Lastly, the organizational dimension (i.e., the aspect that seeks to restructure the larger organization (via decentralization, de-specialization, civilianization, etc.), alter the management style to one that emphasizes the values and culture of the organization over one that stresses formal rules and discipline, and promote accountability via transparent processes and unencumbered information sharing) encompassed all of the tweets that pertained to the dissemination of timely information or included a mention of other units in the department.
Determining which of the eight principles fit within the four themes outlined by Cordner (1995) further advanced the type of analysis that could be conducted on the dataset. The majority of principle assignments to a singular dimension were rather straightforward (e.g., the interaction principle was classified as belonging to the tactical dimension because a main focus was on increasing positive interactions). However, enforcement was also included here as well. This categorization, while potentially not as direct, was included because enforcement tweets also solicited information from the larger community about open cases in their respective precincts. To that extent, by operationalizing Cordner’s (1995) four-dimensional model of community policing, this study is fully grounded in theory and thus adds significantly to the existing calls for theoretically situated community policing research (see Gill et al., 2014).
Content Analysis and Coding
After identifying all the potentially relevant police operational principles (see Heverin & Zach, 2010; IACP, 2014; Kim, Oglesby-Neal, & Mohr, 2017; Lieberman et al., 2017; Mayes, 2020) and thematic community policing dimensions (Cordner, 1995), the content analysis and coding efforts were undertaken by the authors. The coding process involved reading individual tweets and dichotomously indicating (1/yes or 0/no) if a tweet met one or more of the eight police operational principles.9 Tweets could be coded as meeting zero or up to all eight individual principles outlined in the codebook.10 Each of the authors participated in the content analysis and coded an equal subset of randomized tweets from the sample. The first stage of coding tweets involved each author analyzing the same 1,691 tweets (i.e., 10% of the sample) to both establish inter-rater reliability before working through the full sample of tweets11 and to ensure the construct reliability of the eight police operational principles. Kappa coefficients and percent agreement scores were ascertained by comparing the breakdown of tweet scores for each of the eight principles across the three coders.12 After determining that inter-rater reliability was successfully achieved in all eight principles, the remainder of the sample was equally divided across the three coders (4,906 tweets per coder) for phase two.13 Descriptive statistics were run after coding to determine frequencies and modal principles, and to examine preliminary relationships between principles (i.e., when two or more principles were coded as occurring in the same tweet) and themes.
Analytical Technique
We first calculated raw counts and cross tabulations among coded principles and thematic dimensions to address the question of adherence to community policing. These statistics allow us to describe how often tweets contained content that was relevant to the eight pre-defined community policing principles. Additionally, this preliminary analysis would also uncover how often tweets fell within Cordner’s (1995) four thematic dimensions, thereby indicating the community policing principles that are more easily achieved using social media. Next, the cross tabulations provide us with a better understanding of how principles and dimensions are linked, by describing how frequently tweets contain content within certain pairings. With the cross tabulations, we were able to both identify patterns of usage and provide empirical examples of how certain aspects of community policing are more easily blended together. For example, some principles appeared together often (69% of enforcement tweets contained a geo-locational marker), while other principles were not frequently used in tandem (only 2% of enforcement tweets included a mention of a specialty unit).
As an additional summary statistic, we include a metric that represents the average proportion of tweets that contain content in each pair-wise combination of COP principles and themes. This statistic provides added nuance in describing the content of the tweets and gives the reader an understanding of which principles appear to be complementary in nature. For example, NYPD tweet content overwhelmingly included geo-locations, with geo-locations paired with other principles in 44% of all cases. The inclusion of place in the content of messages is an easy but important aspect of virtual COP communications, since using locational information in an online communication medium helps facilitate the connection between virtual and physical spaces.
Conversely, based on the average co-principle measure, it was more difficult to pair “Other Units” or the activities of non-precinct units with content from other principles. This implies the need for greater integration of the activities related to headquarters and centralized units in the everyday communications of precincts for a more cohesive messaging strategy. Second, the adherence statistics reports the number of principles in a tweet. In this case, the more principles and themes a tweet had and the more even the distribution of tweets is across all themes, the more adherent individual tweets and total Twitter usage is to community policing aspects. Such evidence would suggest effectiveness in utilizing Twitter as a strategy to enact community policing. The adherence statistic may have further use as a performance metric whereby over time the number of tweets that have more principles and dimensions should increase from the baseline. The greater proportion of tweets that fall within several principles would be characteristic of a department that is more routinely and effectively promoting COP.
To determine whether there is heterogeneity in the implementation of Twitter across NYPD commands, we employed a series of four logistic regressions.14 Each model specified a new dependent variable that corresponds to whether an individual tweet was coded for a given theme (=1) or not (=0). Each model incorporated the same set of independent variables; these are dichotomous measures that correspond to the patrol borough where the tweet originated. For example, Pi is the probability of a tweet containing the given theme and Xi is whether the tweet originated from a given patrol borough. Therefore, the parameter β0 gives the base log odds (when Xi = 0) and β1 is an estimate of how the odds will differ given Xi = 1, or the patrol borough origination:
Log = log it(Pi) =
Our first model tests the relationship between the dependent variable, that a given tweet will be coded as thematically “philosophical,” and the patrol borough it originated from. For example, should a given theme have a significant association with a patrol borough, and have an Odds Ratio > 1.00, we would say that the patrol borough is predictive of an increased likelihood of generating tweets of that theme. Taken further, this increased likelihood may be indicative of being adherent to COP principles for that theme, whereas a significant decrease in likelihood would mean the opposite. We then conduct three more models the same way to test the relationships between the remaining thematic dimensions (i.e., strategic, tactical, and organizational).
Results
Table 1 reports the raw counts of tweets that were coded as having content that pertains to COP principles and themes. The modal principle was “Geo-coded” tweets with 49% of tweets recording a location. Conversely, only 8% of tweets contained information concerning non-precinct-based units or “Other Units.” The modal theme was “Tactical” with 60% of tweets containing either “Community Interaction” and/or “Enforcement” principles. Conversely, only 27% of tweets fell under the “Organizational” theme pertaining to timely information or non-precinct-based units (i.e., “Other Units”). Finally, very few tweets contained content that could not be ascribed to a policing principle (n = 88, about 0.52% of the overall sample). Similarly, while two principles were the modal tweet category (about 48% of tweets), few tweets contained more than three different principles (roughly 2% of tweets).
Table 1. Counts and Proportions of COP Principles and Thematic Dimensions | |||||||
---|---|---|---|---|---|---|---|
COP Principles | Count | Proportion | Tweet Stats | # of Categories | Count of Tweets | Proportion of Total | |
Enforcement | 2182 | 13% | 0 | 89 | 0.5% | ||
Prevention | 2641 | 16% | 1 | 5581 | 33.0% | ||
Community Interaction | 5604 | 33% | 2 | 8037 | 47.5% | ||
Timely Information | 3209 | 19% | 3 | 2847 | 16.8% | ||
Blue Family | 4657 | 28% | 4 | 344 | 2.0% | ||
Other Units | 1411 | 8% | 5 | 15 | 0.1% | ||
Events | 3648 | 22% | 16913 | 100.0% | |||
Geocoded | 8295 | 49% | |||||
COP Dimensions |
|
| |||||
Philosophical | 7790 | 46% | |||||
Strategic | 7704 | 46% | |||||
Tactical | 10161 | 60% | |||||
Organization | 4504 | 27% |
The cross tabulations in Table 2 show that 69% of tweets with locations also pertained to enforcement actions. Conversely, only 3% of tweets that contained locations pertained to prevention activities, one of the lowest pairings. The lowest pairing of principles were tweets that combined both “Events” and “Enforcement,” which occurred in only 1% of cases. Tweets containing locations were routinely coded for additional principles including “Community Interaction” (51% of all geo-located), “Timely Information” (55%), and “Blue Family” (50%). Table 2 also reports the overlap among thematic dimensions, which displayed much less variation than the principles. Strategic and philosophically themed tweets were the most likely pairing, occurring in 52% of cases containing either theme. In addition, 44% of strategically themed tweets also contained tactical content. However, only 25% of organizationally themed tweets were paired with strategically themed content. Overall, organizationally-themed tweets were seldom paired with other themes.
Table 2. Cross Tabulations - Tweet COP Principles and Thematic Dimensions | |||||||||
---|---|---|---|---|---|---|---|---|---|
COP Principles | Enforcement | Prevention | Community Interaction | Timely Information | Blue Family | Other Units | Events | Geocoded | |
Enforcement | - | 3% | 4% | 16% | 14% | 2% | 1% | 69% | |
Prevention | 3% | - | 10% | 15% | 3% | 2% | 13% | 3% | |
Community Interaction | 4% | 10% | - | 14% | 15% | 6% | 14% | 51% | |
Timely Information | 16% | 15% | 14% | - | 13% | 4% | 22% | 55% | |
Blue Family | 14% | 3% | 15% | 13% | - | 14% | 11% | 50% | |
Other Units | 2% | 2% | 6% | 4% | 14% | - | 14% | 38% | |
Events | 1% | 13% | 14% | 22% | 11% | 14% | - | 45% | |
Geocoded | 69% | 3% | 51% | 55% | 50% | 38% | 45% | - | |
Average Co-Principle | 16% | 7% | 16% | 20% | 17% | 11% | 17% | 44% | |
COP Dimensions | Philosophical | Strategic | Tactical | Organization |
|
|
|
| |
Philosophical | - | 52% | 24% | 23% | |||||
Strategic | 52% | - | 44% | 25% | |||||
Tactical | 24% | 44% | - | 20% | |||||
Organization | 23% | 25% | 20% | - |
|
|
|
| |
Average Co-Theme | 33% | 40% | 29% | 23% |
Logistic Regression Results
Table 3 reports the results from the four logistic regression models. Our first model estimates the likelihood that a tweet will contain philosophically themed content across patrol boroughs with the reference category being the borough of Staten Island. Situating Staten Island as the reference category allows the other patrol boroughs to be more effectively compared, as Staten Island is geographically disconnected relative to the other borough designations.15 Three patrol boroughs report a positive and significant association with the likelihood of posting philosophical tweets relative to Staten Island; these boroughs include Brooklyn South (OR = 1.17), Manhattan South (OR = 1.18), and Queens South (OR = 1.21). The remaining patrol borough designations reported no relationship with philosophical tweets. Model 2 estimated the likelihood of a tweet containing strategic content relative to Staten Island and found that the Bronx (OR = 0.71), Brooklyn North (OR = 0.82), Brooklyn South (OR = 0.84), Manhattan South (OR = 0.73), and Queens North (OR = 0.74, p < 0.001) were all significant predictors of a decreased likelihood relative to Staten Island. Boroughs were relatively consistent with a decreased likelihood in posting strategic tweets; however, there was some variation in significant and non-significant distinctions observed across the patrol boroughs.
Model 3 estimated the likelihood of a tweet containing tactically themed content and also report mixed results. While the Bronx (OR = 1.50, p < 0.001) and Brooklyn North (OR 1.36, p < 0.001) reported positive and significant associations with tactically themed tweet content relative to Staten Island, Manhattan South (OR = 0.71, p < 0.001) reported the opposite relationship. Finally, Model 4 estimated the relationship between organizationally themed content and patrol boroughs but found no significant associations between the two. As such, relative to Staten Island, there was no patrol borough that demonstrated a significantly different level of organizationally themed content, which is suggestive of widespread agreement in the approach to tweeting about other units and the dissemination of timely information.
Table 3. Logistic Regression Results (Odds Ratios) | ||||
---|---|---|---|---|
Measure | Community-Oriented Policing Thematic Dimension | |||
Patrol Borough | Philosophical | Strategic | Tactical | Organizational |
Bronx | 0.95 [0.81, 1.11] | 0.71*** [0.60, 0.83] | 1.50*** [1.28, 1.76] | 0.84 [0.71, 1.01] |
Brooklyn North | 1.17 [0.99, 1.37] | 0.82* [0.69, 0.96] | 1.36*** [1.15, 1.59] | 0.84 [0.70, 1.01] |
Brooklyn South | 1.17* [1.00, 1.37] | 0.84* [0.71, 0.98] | 1.03 [0.88, 1.2] | 1.02 [0.86, 1.22] |
Manhattan North | 1.07 [0.92, 1.24] | 1.05 [0.89, 1.22] | 1.00 [0.86, 1.17] | 1.12 [0.94, 1.32] |
Manhattan South | 1.18* [1.01, 1.38] | 0.73*** [0.62, 0.86] | 0.71*** [0.60, 0.83] | 1.01 [0.84, 1.2] |
Queens North | 0.98 [0.83, 1.15] | 0.74*** [0.63, 0.87] | 1.11 [0.95, 1.3] | 1.02 [0.86, 1.22] |
Queens South | 1.21* [1.02, 1.44] | 0.96 [0.81, 1.14] | 1.02 [0.86, 1.21] | 0.84 [0.69, 1.02] |
Staten Island | Omitted | Omitted | Omitted | Omitted |
N | 15,332 | 15,332 | 15,332 | 15,332 |
Log Likelihood | -10550.49 | -10292.39 | -10457.37 | -8892.656 |
AIC | 21116.97 | 20600.78 | 20930.74 | 17801.31 |
BIC | 21178.07 | 20661.88 | 20991.84 | 17862.41 |
Note: 95% confidence intervals in brackets. AIC = Akaike information criterion, BIC = Bayesian information criterion. Staten Island omitted due to collinearity. | ||||
*p < 0.05, **p < 0.01, ***p < 0.001. |
The final research question examined the potential level of engagement and reach of NYPD tweets by assessing favorites and retweets (see Table 4). This table includes the descriptive statistics that relate to the extent to which NYPD tweet content was shared by the public (retweeted), agreed with or supported by public consumers of the content (favorited), and if the tweet itself was an original tweet or retweet. Tweets that contained content related to “Enforcement” activities were, on average, the most widely shared, with the mean number of retweets for each enforcement-related tweet being nearly 40 times. The proportion of enforcement-related tweets were favorited in similar numbers to the rest of the principles (e.g., 18.75% of tweets with “Enforcement” content), except when compared to the blue family-related tweets, which were often liked and favorited by consumers of the content (e.g., 30.32% of all tweets with “Blue Family” content). Conversely, tweets that contained content related to “Prevention” principles had the least reach. On average, tweets containing “Prevention” content had only 12.41 shares, less than 1/3 of the reach of tweets with “Enforcement” content.
Over half of all tweets that contained “Timely Information” originated from sources outside the NYPD. This suggests that when disseminating “Timely Information,” the NYPD often relies on other institutions and sources to produce the information they use to update the public. Tweets containing “Timely Information” are also relatively widely shared (i.e., on average 33.42 retweets for each tweet), suggestive that this information is being consumed by the public and that the NYPD is providing an important public service. Table 4 provides important information on the reach of NYPD communications, the dissemination of COP principles, and the principles that are most widely consumed and agreed with by the public. Finally, this table reports the degree to which NYPD is reliant upon partners and content from the wider internet to inform the public and displays which principles or aspects of COP that most routinely originate from NYPD and outside sources.
Table 4. Online Community Reach Variables | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
| COP Principles | COP Themes | ||||||||||
Retweet | Enf. | Prev. | Comm. Int. | Timely Info. | Blue Fam. | Other Units | Events | Geo. | Phil. | Strat. | Tact. | Org. |
Obs. | 2,812 | 2,642 | 5,605 | 3,210 | 4,658 | 1,412 | 3,648 | 8,295 | 7,790 | 10,161 | 7,704 | 4,504 |
Mean | 39.90 | 12.41 | 14.15 | 33.42 | 32.99 | 13.34 | 16.45 | 23.36 | 26.03 | 21.74 | 21.24 | 27.15 |
S.D. | 278.09 | 59.39 | 102.65 | 152.38 | 151.46 | 42.23 | 105.07 | 165.57 | 136.60 | 152.60 | 171.96 | 130.61 |
Min | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Max | 10,512 | 2,228 | 4,390 | 4,061 | 3,375 | 1,066 | 4,891 | 10,512 | 4,891 | 10,512 | 10,512 | 4,061 |
Favorite |
|
|
|
|
|
|
|
|
|
|
|
|
Mean | 11.92 | 6.25 | 8.94 | 4.63 | 12.03 | 9.09 | 9.00 | 9.61 | 10.54 | 8.87 | 9.75 | 5.94 |
S.D. | 18.75 | 14.73 | 16.03 | 17.56 | 30.32 | 16.37 | 18.68 | 18.30 | 24.66 | 17.95 | 16.72 | 16.30 |
Min | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Max | 314 | 608 | 503 | 751 | 1174 | 433 | 751 | 503 | 1174 | 608 | 503 | 751 |
Retweet (Y/N) |
|
|
|
|
|
|
|
|
|
|
|
|
Yes | 41.75% | 28.61% | 27.19% | 51.06% | 40.02% | 35.13% | 27.25% | 35.02% | 34.26% | 34.19% | 31.22% | 45.83% |
No | 58.25% | 71.39% | 72.81% | 48.94% | 59.98% | 64.87% | 72.75% | 64.98% | 65.74% | 65.81% | 68.78% | 54.17% |
Discussion and Conclusion
The first research question was concerned with the alignment between NYPD Twitter usage and the principles and themes associated with community policing. The results indicate that Twitter can be used to tweet about all principles and themes of community policing. As opposed to siloed community policing programming or problem/outcome specific interventions and goals, Twitter may be uniquely positioned to contribute to all outlined principles and themes of community policing. Though, there is an observable focus and an enhanced ability to deliver on certain aspects of community policing. Community interaction-based tweets were the most often principle tweeted about (excluding geo-located, which will be discussed later), thus providing evidence that social media may be heavily leaned upon to achieve community engagement and that the assumed goal of Twitter is to promote community connection.
Alternatively, Twitter may be less adept in promoting organizationally themed content surrounding the dissemination of timely information and about the departmental structure and other units. Relatedly, the results suggested that principles and themes were rarely comprehensively tweeted about with few tweets reaching three or more principles. The ability to leverage Twitter as a multi-faceted, within tweet, comprehensive community policing tool was not often readily observed. Outside of the principle’s “Events” and “Timely Information” (and all geo-located comparisons), no other principle pairing was tweeted in conjunction more than 20% of the time. This suggests that the approach to connecting Twitter to community policing may be volume-based and within singular tweets, and that the most effective way to utilize Twitter for community policing is to engage in each principle and theme separately on a tweet specific basis, and to tweet frequently.
The results further suggest that Twitter has the capacity to align with the principles and themes of community policing, though the observed alignment is much more amenable for categories like geo-location, community interaction, and blue family then it is for other units, enforcement, prevention, and timely information. Variation seems to exist in how closely aligned Twitter is across the principles, and many of the multi-principle tweets were propelled by the geo-locational category (with 49% of the overall sample containing reference to a physical location). Interestingly, the only bi-variate analysis where geo-locational tweets was not the highest pairing was for prevention related tweets, which may suggest a lack of precinct or borough-based crime prevention efforts. This is an interesting observation, as many of the other categories and themes reflected a highly localized Twitter focus. The use of geo-locations was especially high, with tweets containing reference to physical locations or specific areas in nearly half of all cases. Community interaction-related content, for reference, was the second most frequently tweeted about theme but was only recorded in 1/3 of all cases with many of these tweets also involving geo-locations (51% contain this pairing).
Considering the level of adherence and specific principles Twitter is able to fulfill within community policing, geo-locational and community interaction principles were most commonly tweeted about. This discovery suggests that Twitter is best positioned to contribute to these aspects of community policing. Though, this may not be the content the public is most interested in engaging with. Community interaction and geo-located tweets may be important community policing tenets in the eyes of police; however, the Twitter audience reached recorded much lower mean interactions with those principles relative to the other categories. Specifically, blue family and enforcement tweets demonstrated the highest average retweets and favorites. This may suggest that a large share of online police engagements are propelled by those associated with policing (e.g., other police accounts) or those with a pre-existing strong pro-police sentiment. If true, this may limit the proposed reach of Twitter in actually connecting the police to a wider audience. This may also reflect the style of police tweets, and how they are self-fulfilling in drawing in a pro-police audience. The inability to connect, dialogue, and create positive relationships with new groups would be a significant shortcoming of Twitter since the paramount goal is to access and connect with an unlimited and new audience. Police may have to change their tweet style, their tweet content, and or their profile identity to reach new groups. The need to alter their Twitter approach may even be exacerbated now following the growing rift between the police and the community following events like the killing of George Floyd and other largely online movements such as “defund the police.”
However, the geo-locational finding is noteworthy and provides support for the unique role social media can play in enacting community policing principles and dimensions. The geo-locational principle involves all references to physical places, which successfully achieves two main community policing objectives. First, referencing physical locations connects the cyber world to the physical world where events, programming, and outcomes are occurring in real-time. This bridges the gap of previously unseen police action and presence by fostering a sense of police visibility in cyberspace. To that extent, this further expands the reach of physically located programming by utilizing social media to take a virtual audience into a physical space for informal facetime (which is a key component of many community policing efforts). This underscores the observed significant relationship between geo-locational tweets referencing locations, and community interaction tweets referencing programming, engagement, and community sponsored initiatives.
Second, the continued reiteration of locations also highlights the organizationally mandated, philosophical, shift to decentralization required in community policing. By organizing the NYPD into precincts and patrol boroughs as opposed to being a singularly defined entity absent operational boundaries, and by allowing individual precincts to run their own social media and outline their boundaries, decentralization is both present and reinforced. In sum, geo-locational tweets are positioned to allow precincts to define the boundaries they oversee and tailor their tweets to specifically focus on such areas and audiences they aspire to connect with.
Remarkably, the only community-oriented policing principle that was under-represented (i.e., less than 10%) was tweets that could be categorized as mentioning “Other Units.” This is not surprising as this grouping was defined as units that exist outside of a precinct (Aviation, Mounted, Emergency Service Unit, etc.) and those that encompassed civilian members of the service (Auxiliary, Crossing Guards, Traffic, etc.). Precincts are less likely to tweet about an outside unit or the accomplishments of a specific outside unit unless they directly relate to an incident that occurred within the confines of said precinct. There were also three principles that occurred less than 20% of the time (enforcement (13%); prevention (16%); and timely info (19%)). These results indicate that patrol boroughs seem to focus more on local problems and initiatives as opposed to posting about citywide issues and general concerns. However, based on the engagement statistics, this may be more reflective of the content that is more desirable to Twitter audiences (e.g., enforcement and timely information).
This local-level focus may have also been the catalyst for the emergence of community interaction tweets, which serves to meet an additional goal of community policing that is predicated on community driven satisfaction, trust, and legitimacy. The volume of community interaction tweets suggests Twitter is an effective medium for fostering, highlighting, and prioritizing community-based programming over exclusively crime-reduction programming and traditional performance metrics. The prioritization of community interaction via Twitter may have also been the driving force behind the emergence of the tactical dimension, which prioritizes local, precinct-based efforts, over the philosophical dimension, which prioritizes large-scale NYPD efforts. As such, individual precincts are more likely to tweet about precinct specific content as opposed to general NYPD content, prioritizing the tactical dimension over the philosophical dimension. The focus on community interaction-based, local-centric content was also observed in Mayes (2020) study on police use of Twitter for image construction across 12 metropolitan police departments.
As it pertains to research question two concerning heterogeneity across patrol boroughs, our regression models find a mixed case for heterogeneity. Agreement was observed in three of four themes, but there was largely an overall lack of agreement across all themes. For instance, there was a consistent lack of significant associations between patrol boroughs and whether a tweet contained organizational content. This suggests that the individual patrol boroughs may be focused on more local activities to the neglect of headquarters-related concerns and events (or activities). For example, an individual precinct, and therefore the resulting patrol borough associated with it, is less likely to tweet about an event being held at headquarters (i.e., 1 Police Plaza (1PP)) unless it specifically pertains to that precinct (for example, someone from that command is being promoted or recognized at a ceremony being held there, etc.). Though, again, this may be positive evidence for the existence of decentralization within the NYPD.
Patrol Borough Manhattan South (PBMS) plays a significant role in decoding the relationships across each of the remaining three dimensions (strategic, tactical, philosophical). Beginning at the tip of lower Manhattan and ending just below Central Park, Manhattan South is a unique entity when compared to the other patrol boroughs. As such, it is home to many NYC landmarks (e.g., One World Trade Center, Wall Street, and the Empire State Building) and features several internationally recognized attractions (e.g., Grand Central Station, Times Square, Broadway, MoMA, Madison Square Garden, etc.) (NYPD, 2017a). Innately, it is the collective of these features that may explain why community-oriented policing is experienced differently in this area.
Compared to PBMS, the Bronx and Brooklyn North patrol boroughs are significantly more likely to tweet about “Community Interaction” and “Enforcement” related issues which focus on residential populations (e.g., the tactical dimension), whereas Manhattan South is less likely to focus on such material. Manhattan South is also predictive of an increased likelihood of communicating philosophical content (e.g., this include tweets related to “Causes/Events/Awareness” and “Blue Family”), which generally underscore larger departmental initiatives since headquarters and admin are both located there. This suggests that while not as focused on local crime problems in terms of their tweet content, they are helping to drive other community policing principles and large-scale NYPD thematic philosophies and dimensions. Taken together, these two notable differences suggest that the patrol borough of Manhattan South is significantly and philosophically different than the other patrol boroughs, while the other patrol boroughs often differ in how they operate based on nuances across local contexts. This provides evidence that the social media strategy is heterogenous across patrol boroughs and is locally focused. This degree heterogeneity was also observed across German police departments on a national scale, which further indicates that there are observable differences in the way police tend to tweet based on the communities they serve (Jungblut & Jungblut, 2021).
The localized use of Twitter is likely linked to aspects of decentralization, but the NYPD and other implementing agencies must be careful to not rely solely on local efforts at the expense of larger philosophical concerns (like mission statements). Although the application of Twitter may be best suited for promoting locally driven, community specific content, the core mission of social media engagement must be uniform as to not fracture larger departmental standards. Social media is uniquely positioned to contribute to each aspect of community policing, though, and the evidence of its effectiveness in fostering several key principles and dimensions is apparent. Social media can also be rolled out without the associated costs common to other programming and can provide quick access to points to connect the police and community. Notably, the effective application of Twitter can directly result in reinforcing organizational shifts like decentralization, developing community partnerships, and leveraging the real-time orientation of cyberspace to connect to the physical world.
However, social media is not a catchall program or solution to all community policing related endeavors. For instance, some aspects of problem solving, one of three major community policing components, may be better served through another intervention, program, or strategy. This inference is supportive of early research that suggests police should attempt to integrate a wide range of programmatic responses to address specifically identified issues (i.e., as opposed to using a single hammer to hit every nail (Sparrow, 2015)) by simply focusing on multiple, small-scale problems and harm reduction efforts (Goldstein, 1979; Eck & Spelman, 1987).
Moreover, as it relates to the last research question, the engagement metrics reveal some interesting trends and patterns regarding police and community relations. The principles and themes police are most likely to tweet about do not necessarily reflect the content that the public is interested in interacting with the most. This disconnect may reflect a chasm between police and public goals for Twitter and community engagement. Moreover, it may be evidence that the online context of Twitter is not being effectively leveraged for police to reach new audiences. Despite a lower volume of tweets, enforcement specific tweets (i.e., those that most often focused on apprehension efforts and sought investigative help) were some of the most highly interacted with online. This is an interesting finding as it is not central to many community policing efforts.
Limitations
This study provided an exploratory look at the relationship between social media use and community policing principles. It provided insights into how the NYPD uses Twitter to accomplish community policing; however, it does not consider whether the use of social media as an application of community policing has any effect on crime levels. A central aim of any policing program, including those associated with community policing, is to achieve certain deliverables, which generally includes crime reduction. To address this limitation, future studies should seek to explore how the use of social media meets such criteria. For example, research could consider how a patrol borough or a department uses Twitter and how that relates to observed crime trends.
It is also important to note that this study may not be as generalizable as intended because of the uniqueness of the NYPD.16 The level of analysis utilized was meant to not only consider the different characteristics that make up each of the individual precincts but also the differences that exist between patrol boroughs. The application of social media and its alignment with community policing principles may differ in departments that operate with far fewer jurisdictions and Twitter accounts.
Another limitation of this study is that it only examines one year of data. Although, the yearly tweet volume and sample size was adequately high and representative, the findings may be different after the inclusion of multiple years. Furthermore, by the end of 2017, the Deputy Commissioner of Strategic Communications was able to publish and distribute the NYPD Social Media Guide and a new position within the commands was created (i.e., the digital communications officers (DCOs)) to monitor their precinct’s social media account(s). This agency wide policy change may have impacted the application and use of social media at the precinct-level following implementation. Future research should consider the potential for change in NYPD Twitter usage following the publication of the social media guide, and whether the guide aligns with community policing principles.
References
Bayley, D. H. (1994). Police for the future. New York: Oxford.
Brainard, L., & Edlins, M. (2015). Top 10 municipal police departments and their social media usage. American Review of Public Administration, 45(6), 728-745.
Bratton, W. J. (1996, Spring). New strategies for combatting crime in New York City. Fordham Urban Law Journal, 23, 781-795.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20(1), 37-46.
Cordner, G. W. (1995). Community policing: Elements and effects. In R. G. Dunham, & G. P. Alpert (Eds.), Critical issues in policing, (5th ed.) (pp. 401-418). Long Grove, IL: Waveland Press, Inc.
Crabtree, S. (2020, July 22). Most Americans say policing needs ‘major changes.’ Gallup. Retrieved December 14, 2020, from https://news.gallup.com/poll/315962/americans-say-policing-needs-major-changes.aspx
Criado, J. I., Sandoval-Almazan, R., & Gil-Garcia, J. R. (2013). Government innovation through social media. Government Information Quarterly, 30, 319-326.
Eck, J., & Spelman, W. (1987). Who ya gonna call? The police as problem-busters. Crime & Delinquency, 33(1), 31-52.
Forsey, C. (2019, January 29). What is Twitter and how does it work? HubSpot. Retrieved December 14, 2020, from https://blog.hubspot.com/marketing/what-is-Twitter
Gill, C., Weisburd, D., Telep, C. W., Vitter, Z., & Bennett, T. (2014). Community-oriented policing to reduce crime, disorder and fear and increase satisfaction and legitimacy among citizens: A systematic review. Journal of Experimental Criminology, 10, 399-428.
Goldstein, H. (1979). Improving Policing: A Problem-Oriented Approach. Crime & Delinquency, 25(2), 236–258.
Heverin, T., & Zach, L. (2010). Twitter for city police department information sharing. Proceedings of the Association for Information Science and Technology, 47(1), 1-7.
International Association of Chiefs of Police. (2014). 2014 social media survey results. Retrieved on July 19, 2020, from https://s3-eu-west-1.amazonaws.com/s3-euw1-ap-pe-ws4-cws-documents.ri-prod/9780323357722/Ch7/7-01-IACP-2014-Survey-Results.pdf
Jungblut, M., & Jungblut, J. (2021). Do organizational differences matter for the use of social media by public organizations? A computational analysis of the way the German police use Twitter for external communication. Public Administration, 1-20. DOI: 10.1111/padm.12747
Kim, K., Oglesby-Neal, A., & Mohr, E. (2017). 2016 Law enforcement use of social media survey. URBAN Institute. Retrieved December 12, 2017, from http://askus.library.wwu.edu/faq/119191
Lieberman, J. D., Koetzle, D., & Sakiyama, M. (2013). Police departments’ use of Facebook: Patterns and policy issues. Police Quarterly, 16(4), 438-462.
Maguire, E. R., & Mastrofski, S. D. (2000). Patterns of community policing in the United States. Police Quarterly, 3(1), 4-45.
Mastrofski, S. (2006). Community policing: A skeptical view. In D. Weisburd, & A. A. Braga (Eds.), Police innovation: Contrasting perspectives (pp. 44-73). Cambridge: Cambridge University Press.
Mateescu, A., Brunton, D., Rosenblat, A., Patton, D., Gold, Z., & Boyd, D. (2015). Social media surveillance and law enforcement. Data & Civil Rights: A New Era of Policing and Justice, 27, 1-11.
Maxfield, M. G., & Babbie, E. R. (2015). Research methods for criminal justice and criminology (7th ed.). Stamford, CT: Cengage Learning.
Mayes, L. (2020). Social media and community-oriented policing: Examining the organizational image construction of municipal police on Twitter and Facebook. Police Practice and Research: An International Journal, 1-18.
McHugh, M. (2012). Interrater reliability: The kappa statistic. Biochemia Medica, 22(3), 276-282.
Meijer, A., & Thaens, M. (2013). Social media strategies: Understanding the differences between North American police departments. Government Information Quarterly, 30, 343-350.
Meijer, A. J., & Torenvlied, R. (2016). Social media and the new organization of government communications: An empirical analysis of Twitter usage by the Dutch police. American Review of Public Administration, 46(2), 143-161.
Mergel, I. (2012). The social media innovation challenge in the public sector. Information Polity, 17, 281-292.
Nagy, A. R. & Podolny, J. (2008). William Bratton and the NYPD. Yale School of Management, Yale Case 07-015.
NYPD. (2016). The Police Commissioner’s report. Spring 3100, pp. 1-80. Retrieved January 9, 2020, from https://www1.nyc.gov/assets/nypd/downloads/pdf/publications/pc-report-2016.pdf.
NYPD. (2017a). The Police Commissioner’s report. Spring 3100, pp. 1-96. Retrieved February 7, 2020, from https://www1.nyc.gov/assets/nypd/downloads/pdf/publications/pc-report-2017.pdf
NYPD. (2017b). Social media guide. New York: NYPD, The Office of the Deputy Commissioner, Strategic Communications.
O’Connor, C. D. (2017). The police on Twitter: Image management, community building, and implications for policing in Canada. Policing and Society, 27(8), 899-912.
President’s Task Force on 21st Century Policing. (2015). Final report of the President’s task force on 21st century policing. Washington, DC: Office of Community Oriented Policing Services.
Rukus, J., Warner, M. & Zhang, X. (2018). Community policing: Least effective where need is greatest. Crime & Delinquency, 64(14), 1858-1881.
Skogan, W. G. (2006). The promise of community policing. In D. Weisburd, & A. A. Braga (Eds.), Police innovation: Contrasting perspectives (pp. 27-43). Cambridge: Cambridge University Press.
Skolnick, J. H., & Bayley, D. H. (1988). Theme and variation in community policing. In M. Tonry, & N. Morris (Eds.), Institutional environments and organizations (pp. 100-112). Thousand Oaks, CA: Sage.
Sparks, D. (2017, April 27). How many users does Twitter have? The Motley Fool. Retrieved June 23, 2020, from https://www.fool.com/investing/2017/04/27/how-many-users-does-Twitter-have.aspx
Sparrow, M. (2015). Measuring performance in a modern police organization. Psychological Issues in Human Resource Management, 3(2), 17-52.
Telep, C. W., & Weisburd, D. (2016). Policing. In D. Weisburd, D. P. Farrington, & C. Gill (Eds.), What works in crime prevention and rehabilitation (pp. 137-168). New York, NY: Springer.
Uchida, C. D. (2005). The development of the American police: An historical overview. In R. G. Dunham, & G. P. Alpert (Eds.), Critical issues in policing, (5th ed.) (pp. 20-40). Long Grove, IL: Waveland Press, Inc.
Walsh, J. P. (2020). Social media and border security: Twitter use by migration policing agencies. Policing and Society: An International Journal of Research and Policy, 30(10), 1138-1156.
Williams, C. B., Fedorowicz, J., Kavanaugh, A., Mentzer, K., Thatcher, J. B., & Xu, J. (2018). Leveraging social media to achieve a community policing agenda. Government Information Quarterly, 35, 210-222.
Zhao, J., Scheider, M. C., & Thurman, Q. (2002). Funding community policing to reduce crime: Have COPS grants made a difference? Criminology & Public Policy, 2(1), 7-3