Social media engagement tactics in U.S. community policing: Potential privacy and security concerns

Abstract

Despite law enforcement’s best efforts to use social media as a means of community policing, some engagement tactics may lead citizens to disclose personally identifiable information (PII). We coded 200 tweets with the popular #9PMRoutine that tagged @PascoSheriff (Florida) for participant PII. We found numerous postings of adults’ and children’s PII that are problematic including pictures, health information and security-related comments about their routines or vacations. Implications for law enforcement to protect their communities are discussed as well as opportunities to continue to cultivate their online relationships in a more secure forum. We also provide future research directions.

Keywords

Community-oriented policing social networking privacy

Introduction

Community and law enforcement relations are important in American society. Michael Brown’s death in August 2014 when white Ferguson, Missouri police officer Darren Wilson shot the unarmed African-American man, however, brought community interactions to the forefront as these relationships deteriorated (Chanin and Espinosa, 2016). In response, law enforcement agencies (LEAs) looked to technological solutions ranging from officer-worn body cameras to enhance transparency and accountability (Schneider, 2017) to more effective use of their digital communication strategy (Brainard, 2016; Walsh and O’Conner, 2019). In particular, an effective presence on social media platforms became necessary (Aiello, 2018; Brainard and Edlins, 2015; Huang et al ., 2017) as it ‘can enhance existing [community policing] efforts by bringing more residents into a relationship with police, allowing for asynchronous communication’ (Brainard, 2016: 119). According to the International Association for Chiefs of Police (IACP), 81.1% of agencies used social media in 2010 with 40.6% using it for community outreach/citizen engagement. By Fall 2015, that number had climbed to 96.4% using social media with 83.4% reporting that they used it for community outreach/citizen engagement (IACP, 2010, 2015). Most early efforts focused on using social media to aid in investigations and provide important notifications or positive stories to the public and media (Hanson, 2011) rather than tapping into social media’s power of engagement and relationship development. Consequently, many LEAs were not engaging in community building as effectively as they could have been (Huang et al ., 2017). As the International Association of Chiefs of Police (IACP, n.d.) notes in its social media for law enforcement Twitter guidance: ‘Engagement is your goal. Meaningful interaction matters, as it builds loyalty and trust. Twitter is the ideal platform for engagement’.

As a means of prompting consumer engagement, advertisers/brands use a variety of tactics such as Twitter chats, quizzes, or asking questions, some of which include posting pictures in response, and often stimulated by the opportunity to win prizes (Fox and Hoy, 2019). These efforts are intended to develop relationships between the sponsor account and its followers. LEAs are turning to such marketing tactics to generate engagement as part of their community policing efforts.

These engagement tactics, however, may result in participants sharing personally identifiable information (PII) about themselves and others, including children. Emerging research that looks at mothers posting information about their children in similar scenarios suggests that this behavior may put their children’s privacy, and potentially safety, at-risk (Fox and Hoy, 2019). Similarly, we wondered if the same behavior might be happening as citizens engaged with police social media accounts. Specifically, this study explores the extent to which Twitter users may be disclosing PII or other sensitive information in response to a law enforcement social media engagement tactic and the resulting potential privacy and safety concerns. To investigate this issue, we conducted a content analysis of 200 tweets taken from one day each in January, June and October 2019 involving a popular engagement tactic, the #9PMRoutine, instituted by the Pasco County (Florida) Sheriff’s Office (Pasco County) but also adopted by many LEAs across the United States.

We begin with Pasco County’s background and the development, use of, and successes with social media as a community policing tool based on an interview with its Assistant Executive Director, Chase Daniels (2019). Then we provide a literature review of community policing and social media use and online privacy as it relates to our study. Next, we describe our content analysis method, including the use of the SalesForce social listening tool, Social Studio. Based on our results we offer guidance for law enforcement organizations’ social media strategists and administrators. We conclude with directions for future research.

Pasco County Sheriff’s Office Social Media and the #9PMRoutine

Pasco County, Florida is a bedroom community of Tampa with a population of around 520,000. Pasco County had been on Facebook and Twitter since 2009 and has an Instagram account that went active in late 2016. In June 2016, the sheriff tasked Daniels with reinventing the organization’s social media presence from one that primarily reported ‘bad news’ such as a road closure or announcing an arrest. Drawing upon conversations with organizations such as the Tampa Bay Rays baseball team and Disney World, Daniels sought insight into how ‘to create some type of engaged following for whatever brand it is’ (Daniels, 2019). After a slow, gradual build, Pasco County now boasts one of the largest social media followings in law enforcement in Florida.

Daniels says that his primary goal for its social media is what he calls the ‘IEP System—Inform, Engage, Promote’. The agency focuses on its mission to inform the public, similar to typical press conferences. However, he notes the importance of ‘engagement, bringing people back, making people feel like they are a part of something’. A large part of its engagement efforts’ success is founded in #9PMRoutine.

The #9PMRoutine

In August 2016, shortly after Daniels began the social media revamp, he and the sheriff discussed how they could create a call to action in their social media that would get citizens talking about their accounts. The conversation included a recent trend of unlocked auto burglaries. They brainstormed how they could create a call to action that would remind people to lock their doors and cars. The ‘9 PM Routine’ was the result as it reflected when most people were coming home after youth sports or dinner.

Daniels noted that after a week of #9PMRoutine they considered abandoning it as it ‘wasn’t taking off’. However, a few days later ‘we got a tweet from someone in Canada, “I have locked my doors”’. From there, participation exploded. The organization introduced the idea of coloring in a worldwide map green when someone from a certain state or country tags the Pasco County and includes #9PMRoutine identifying his/her state or country. Throughout the evening the map is updated to show locations that have checked in as indicated by the state or country going green. The account further encourages followers from remaining locations to ‘get the map green’ creating a sense of global team camaraderie each night. As testimony to the #9PMRoutine success, Pasco County’s Twitter account has gone from 40,000 in June 2016 to 132,500 followers in August 2020. Daniels notes, ‘since we introduced the 9 PM routine unlocked auto burglaries in Pasco County are down about 35 percent’. More importantly, by developing this relationship with its constituency, Pasco County has the platforms and audience in place to amplify messaging during a crisis situation.

Literature review

Community policing & social media use

Community policing is defined as ‘a law enforcement philosophy that allows officers to continuously operate in the same area to create a stronger bond with the citizens living and working in that area’ (Lortz, 2016). This approach gained traction in the 1980s to make law enforcement efforts less ‘reactive, command and control organizations’ and more proactive in social problem solving within their communities (Brainard and Edlins, 2015: 729). Community participation is key to effective community policing (Brainard and Edlins, 2015). The advent of social media has provided a new mechanism by which to facilitate this participation. In their review of the literature on social media and policing, Walsh and O’Conner (2019) identified two broad purposes of social media for law enforcement agencies: surveillance and organizational communication. Our research focuses on this latter use.

Social media provides law enforcement agencies a means to bypass traditional mass media to communicate with and engage the public in social interaction (Meijer and Thaens, 2013; Walsh and O’Conner, 2019). In doing so, it allows them to get out notifications in a timely manner for both ‘mundane issues such as traffic updates’ to those related to a crisis situation such as active shooters or natural disasters (Walsh and O’Conner, 2019: 5). The authors observe that ‘while less common, social media are also utilized as a tool of bidirectional communication and contextualized as community policing’ (p. 6).

The majority of prior research conducts content analysis, either quantitative or thematic, asking the same basic question: How are LEAs using social media? These studies included profiles of platform usage to type of content posted to the level of citizen interaction (see Aiello, 2018; Brainard and Edlins, 2015; Huang et al ., 2017; Huang and Wu, 2018; Williams et al ., 2018).

Huang et al . (2017) classified Twitter users who were either mentioned by LEAs or mentioned the LEAs in their posts. They found relationships between the type of stakeholder and subsequent sentiment of the stakeholders’ posts. In later research, Huang and Wu (2018) coded tweets into three categories of social media strategies: push, networking, and pull. Push included content related to crime, traffic, and announcements. Networking provided information, appreciation, tips, and personnel-related content. Pull tweets were requests. The primary strategy used was push with 78.3% of tweets. Networking represented only 8% of total tweets. However, Huang and Wu (2018) found that networking strategies generated significantly more favorites than the other two strategies. Networking’s ability to tap into emotional connections may explain this finding (Huang and Wu, 2018). Williamson et al . (2018) noted that those who make an effort to engage with an LEA tweet may be more prone to already having a positive disposition toward the LEA.

Thus, prior studies have provided thorough profiles of what LEAs are posting on which social media platforms, how followers might react and how successful their social media strategies may be. What’s missing from this research, however, is an examination of the personal information that citizens may be sharing in response to LEA posts which may put their online privacy, and potentially physical security, at risk.

U.S. online privacy considerations

Protecting individuals’ privacy and personally identifiable information (PII) as they engage with businesses is a key responsibility of the U.S. Federal Trade Commission (FTC, 2018). With particular focus on children, the Children’s Online Privacy Protection Act (COPPA), passed by Congress in 1998 and updated in 2012, seeks to give parents control over their children’s PII (FTC, 2013).

According to the U.S. General Services Administration (2014), PII for adults and children is defined as

information that can be used to distinguish or trace an individual’s identity, either alone or when combined with other personal or identifying information that is linked or linkable to a specific individual. The definition of PII is not anchored to any single category of information or technology. Rather, it requires a case-by-case assessment of the specific risk that an individual can be identified. In performing this assessment, an agency needs to recognize that non-PII can become PII whenever additional information is made publicly available—in any medium and from any source—that, when combined with other available information, could be used to identify an individual.

For children, the FTC (2015) specifically states that PII includes information such as first and last name; home or other physical address including street name and name of a city or town; online contact information; screen/user name that functions as online contact information; telephone number; social security number; geolocation information that is sufficient to identify street name and name of city or town; and photographs, video or audio files that contain an individual’s image or voice.

Around 2000, the FTC conducted a series of ‘privacy sweeps’ to assess the extent to which commercial websites were collecting PII (e.g. FTC 1998, 2000). This work was extended to the non-commercial sector by examining church websites (Hoy and Phelps, 2003) and the top 100 non-profit organization websites (Hoy and Phelps, 2009). The authors found that nearly all of the churches and non-profit organizations collected some form of PII from both adults and children. These organizations also posted adult and child PII on their websites, and in some cases, health-related information. The church websites also posted information that included an individual’s location (e.g. being out of town) or potential vulnerability (e.g. elderly shut-in). Hoy and Phelps (2003) note that this type of shared information puts not only an individual’s privacy, but also personal safety at risk.

Fox and Hoy (2019) took a different perspective regarding how children’s online privacy and safety might be compromised. Rather than coding for PII that an organization might be collecting directly from children, thus triggering compliance with COPPA, they studied how young mothers shared their children’s PII during a Twitter chat with a popular children’s clothing brand. The mother’s account had to be set to public to participate, allowing anyone access. They found that these women readily shared pictures, stories, and other PII about their children during this engagement tactic.

As LEAs turn to social media engagement tactics to enhance community relations, we similarly question whether citizens are posting their own or others’ PII in response. Specifically:

RQ1: To what extent do participants in the #9PMRoutine who tag @PascoSheriff post PII?

The U.S. Health Insurance Portability and Accountability Act of 1996 (HIPAA) provides citizens’ rights regarding their health information. Although HIPAA’s focus is on health care providers and insurers, individuals are encouraged to know who has seen their health information (U.S. Department of Health & Human Services (DHHS), n.d.). Yin et al . (2015) found that individuals readily tweet about their own and others’ personal health status for a variety of conditions ranging from the more mundane such as migraines and allergies to those that are quite sensitive such as mental health and cancer.

RQ2: To what extent do participants in the #9PMRoutine who tag @PascoSheriff post health-related information?

Social media has been dubbed ‘the best free tool for burglars’ (GetSafe.com, n.d.) Not only can posting personal data enhance the vulnerability to identity theft, but comments about one’s travels or routines may signal when a home is vacant. In their research among individuals convicted of car theft, Quinn and Grove (2018) identified situational cues used to assist in the burglary. Once offenders had identified a suitable location and vehicle, the next step was to check if the door was unlocked. Zawisza and Garza (2017: 204) noted that ‘at the point of entry, dogs, security systems, and locks are three of the most common stimuli that affect the decision to burglarize’. Locks and deadbolts don’t always deter offenders ‘since they can be overcome with skill and ingenuity’ (Zawisza and Garza, 2017: 204). The authors found that study participants used house aesthetics (i.e. ‘fancy, big houses’ or ‘run-down houses’) to determine vulnerability to burglary. Conceivably, the information that citizens post in response to #9PMRoutine might contribute to a criminal assessing potential vulnerability.

RQ3: To what extent do participants in the #9PMRoutine who tag @PascoSheriff post non-PII information that might put their own or others’ safety and security at risk?

Finally, it is important when examining these privacy considerations to understand that these posts were not private discussions between Twitter users and Pasco County Sheriff’s Office. To examine the extent to which these possible security concerns were visible to a wider public audience, one must also consider the engagement these posts generated among people outside of Pasco County.

RQ4: To what extent are the posts by the participants in the #9PMRoutine tagging @PascoSheriff liked and retweeted by @PascoSheriff and other users?

Method

Content analysis provides a method by to which to systematically, objectively and quantitatively study and analyze communication (Kerlinger, 1986). One of its primary uses is to describe communication content (Wimmer and Dominick, 1997), the application within this study.

Sampling frame

The researchers utilized Salesforce Radian6 to pull the study sample. Radian6 allows keyword searches of publicly available social media posts, along with filtering using hashtag and the tagging of a profile through a feature called ‘Topic Profiles’. These topic profiles allow the software to perform complex searches of social media platforms using keyword searches through the platform APIs.

The researchers examined data from January 25, June 21 and October 22, 2019. These dates represented ‘normal days’ without large events or campaigns by the Pasco County that would interfere with the sample. These dates also reflected the department’s participation in the popular reality television series Live PD (January 25) and when they did not (June and October) to see if agency participation influenced inclusion of PII in respondents’ posts.

To ensure that news about local crime in Pasco County did not have an effect on the sample, we examined news stories aggregated by Google. Using the term ‘“Pasco County” Crime’ we examined all news stories aggregated by Google during a 7-day time period leading to the day of our sampled Tweets. In total, there were two crimes directly related to the mission of #9PMRoutine: a peeping Tom on the day of one of the samples, and a burglary 1 week before one of the samples. Neither of the stories received widespread coverage online. Furthermore, the coders had an ‘other comments’ category on their code sheets where they noted anything that was not on the code sheet, including references to crimes. None of the crimes that were committed during the week leading up to each sample were referenced in any of the sampled tweets. Therefore the randomly drawn samples across the three time periods were representative of typical citizen engagement with #9PMRoutine.

The researchers created a ‘Topic Profile’, allowing them to pull posts that utilized #9PMRoutine and tagged @PascoSheriff. The software allowed the researchers to filter out any tweets that were retweets with no added content. Social Studio provided a download of the tweets in a .CSV file. Utilizing Microsoft Excel’s filtering, the researchers eliminated any posts authored by @PascoSheriff.

Using Microsoft Excel’s formulas, the researchers assigned randomly generated numbers to each tweet, and sorted from smallest to largest. Using the filters in Excel, the researchers eliminated any posts authored by a law enforcement agency. See Figure 1.

Figure 1.

Sample refinement.

Coding scheme development

We developed a standardized coding sheet using Hoy and Phelps’ work (2003, 2009) and modified it to reflect literature related to community policing. We coded for the presence of PII for adults and children (RQ1). Specifically, we coded adult name, child name, adult state, child state, adult city/county, and child city/county), adult photographs and child photographs. We coded for health-related information (RQ2). To address RQ3 we coded for potential security risks such as photographs of an identifiable home interior, pet information and information that could suggest a routine for when they are not at home. For RQ4, we coded for post engagement in the form of @PascoSheriff Likes and @PascoSheriff Retweets. We also coded for engagement (likes and retweets) from non-Pasco Sheriff accounts.

Coder training and procedure

This study utilized two coders to ensure code reliability. The coders worked with the third researcher to discuss the code sheet and determine rules to ensure all parties were aligned in thought and process. Fifteen tweets from outside of the sample were coded by the coders separately to test the code sheet. The coders met and discussed the code sheet further before beginning the process of coding the sample. The first 100 tweets (from January 25, 2019) were broken into five sets, and after sample refinement, a sixth ‘alternate’ set was examined as well. The coders coded each set separately and met after each was complete to discuss the code sheet and resolve differences. The next 50 tweets (from June 21, 2019) were broken up into two batches. Mirroring the process in the first batches, the coders met after each batch to see what differences needed to be resolved. The researchers used Cohen’s Kappa to determine intercoder reliability, the results can be seen in Table 1. After determining that their agreement level was good enough to move forward, the coders divided the remaining 50 and individually coded.

Table 1.

Intercoder reliability tests (n = 150).

Item	Cohen’s Kappa
Adult Name	.857
Adult State	.882
Adult City/County	.955
Adult Photograph/Video	1.000
Adult Age	1.000
Child Name	1.000
Child State	1.000
Child City/County	1.000
Child Photograph/Video	1.000
Child Age	1.000
Health Information Disclosed	1.000
Identifiable Home Interior	1.000
Pet Information/Photograph	.950
Door Lock Shown	1.000

Comparison across samples

Before addressing the research questions, we ran an analysis of variance to examine potential differences across the three data collection points. The first 100 of the sample occurred during a period when Pasco County Sheriff was on the television show ‘Live PD’ and the second 100 occurred after they left the show. To determine if being on Live PD affected whether people shared PII, the researchers conducted a one-way ANOVA examining the variables ‘Was Pasco County Sheriff’s office on Live PD—yes/no’ and the number of PII disclosures present. We found that there was no significant differences between the means of PII disclosures based on whether Pasco County Sheriff’s Office was on Live PD at the time F(1, 198) = 1.004, p = .318.

Because our samples were pulled from different days of the week, we wanted to ensure there were no significant differences based on days of the week. There was not a significant difference in the means of PII disclosures based on the day of the week F(1, 198) = 2.775, p = .097. However, because of the near significance of day of the week, further examination was conducted. This examination showed that on Fridays, the @PascoSheriff Twitter account is more active in pushing their ‘Paint the map green’ campaign that actively asks for states and gamifies the process. Therefore we ran a one way ANOVA examining the means of PII disclosures ignoring the presence of ‘adult state’, the difference was not significant F(1,198) = 2.190, p = .140. Thus, we concluded that the three samples were comparable with no significant differences. Therefore, we addressed the four research questions based on the aggregate data of 200 tweets. Table 2 depicts the results for these research questions.

Table 2.

Frequency and percentage of PII occurrences (n = 200).

Items	Number of Tweets	Percent of Tweets
RQ1: Extent to which participants posted personally identifiable information (PII)
Any PII	186	93%
Any Adult PII	185	92.5%
Adult Name	149	74.5%
Adult State	143	71.5%
Adult City/County	61	30.5%
Adult Photograph/Video	1	.55
Adult Age	0	0%
Any Child PII	6	3%
Child Name	0	0%
Child State	3	1.5%
Child City/County	1	.5%
Child Photograph/Video	4	2%
Child Age	2	1%
RQ2: Extent to which participants posted health-related information
Health Information Disclosed	5	2.5%
RQ3: Extent to which participants posted information that might puttheir own or others’ safety and security at risk
Any Safety/Security Risk	38	19%
Identifiable Home Interior Shown	2	1%
Pet Information/Photograph	31	15.5%
Routine Reference	8	4%
Door Lock Shown	1	.5%
Miscellaneous	1	.5%
RQ4: Extent to which participants’ posts were engaged with by @PascoSheriff or other accounts
@PascoSheriff likes	116	58%
@PascoSheriff Retweets	0	0%
Other Account Total Likes	429	N/A
Participant tweets liked by other accounts	93	46.5%
Other Account Total Retweets	23	N/A
Participant tweets retweeted by other accounts	13	6.5%

Personally identifiable information posted

RQ1 asked to what extent did participants in the #9PMRoutine provide PII about themselves or others. 186 of the 200 tweets (93%) disclosed 411 instances of PII. Most PII disclosures (n = 292, 71%) were either present in the profile to begin with (name) or were general geographic information provided to participate in Pasco County’s gamification of the #9PMRoutine event (i.e. the state). However, there were still many individual disclosures that were classified as problematic (n = 119, 29%).

Concerning children’s PII, three of the 200 posts included photographs, and one post included a video. One post showed a child’s picture and the state in which s/he resides. The picture also contained an article of clothing associated with a community sports team that could be used to discover the location of the family. The name of a guardian was explicitly present in the tweet, and the occupation was present in the profile bio which was visible in the background upon clicking the tweet.

Another tweet disclosed that a grandparent was watching the two grandchildren, as well as the city and state of their home were disclosed in the post. Another post featured multiple images of a child, one of which was at a public event with another child present who does not seem to be related to the user who posted the information. This post also featured the area in which the event took place.

Health, safety and security-related information posted

RQs 2 and 3 examined the extent to which health and safety/security-related information was posted respectively. Five posts, 2.5% of total tweets, disclosed health information. In one post a user disclosed an ankle injury and included the state and a generic region within that state. Others talked about seasonal sicknesses such as having the flu, or normal occurrences such as a migraine.

A total of 38 posts, 19% of total tweets, had some form of safety or security-related information (RQ3). The most common form was to share pet-related information or photographs (31 posts or 15.5% of total tweets) followed by references to routines (8 posts or 4% of total tweets). Pet-related posts could include anything from a generic picture of a pet, or an image of the pet with the pet’s name included to more problematic issues such as a pet’s image within identifiable home interiors. Routine references ranged from discussing birthdays and graduation dates to mentions the user was on vacation in a different state or future travel. One notable example included an individual disclosing that they were in another state, providing their full name, the city and state they were vacationing in and an image of their pet who was with them. Five of the twelve security disclosures that were not of pets were disclosures present in pet-related images, for example, a home interior image with a dog in the foreground.

Multiple people provided information about either crimes attempted/committed against them or concerns about neighborhood safety. For example, a user posted about an encounter of someone allegedly stealing their keys. While their location is not explicitly posted in the tweet (and thus not coded as a disclosure) their state and region within that state is present in their bio.

Extent of engagement with participants’ posts

RQ4 assessed the extent to which both @PascoSheriff and other individuals engage with participants’ posts. To do so, every tweet that was analyzed directly tagged @PascoSheriff. As stated before, Social Studio can only pull information from profiles where the privacy settings are set to ‘public’. Subsequently, every post had the potential to be viewed by any user not directly blocked by the author’s account, even if the account were blocked by @PascoSheriff. Furthermore, individuals without Twitter accounts could also see these public posts.

As shown in Table 2, the @PascoSheriff account engaged with 58% (116 posts) of the participants via ‘liking’ the posts. Notably, @PascoSheriff did not retweet any of the participants’ posts in which it was tagged. Of the 200 posts, 93 (46.5%) received at least one ‘like’ from a non-PascoSheriff account. In total, there were 429 ‘likes’ across the sample from non-PascoSheriff accounts. Furthermore, all 23 retweets were from other accounts. These 23 retweets were spread across 13 of the 200 posts (6.5%). While we could not code for whether tweets were from followers of the accounts, these tweets were available for public viewing beyond followers of the user and @PascoSheriff.

Discussion

Community policing meets privacy disclosures

To the authors’ knowledge, this is the first study to explore how community policing’s use of social media engagement tactics might result in citizens sharing PII that might put their own and others’, including children’s, privacy and security at risk. Nearly all the participants’ responding tweets in our sample included some form of PII. While we acknowledge that the only information that participation in the #9PMRoutine is requesting is one’s state, most participants by default have their name associated with their Twitter accounts. Many also provided a range of PII about themselves, and in some cases, children. They also discuss health information, show images of their pets and home interiors as well as information about their routines or going on vacation.

Although only a few posts contained children’s PII or health-related information, because this was a random sample from the participating tweet population, it is possible to weight the sample’s findings to see how a month’s worth of data may look. According to Social Studio, there were 7,500 posts containing #9PMRoutine and tagging @PascoSheriff that were not retweets in 30 days from November 1 to December 1, 2019. Assuming the refinement of the population to the Sample Frame holds up, then about 4,650 posts would be in line with our final sample. This would predict that each month there are over 2,700 problematic personal disclosures made each month using #9PMRoutine. This extrapolation includes over 90 images or videos of children and over 880 instances of security disclosures that may put the Twitter user’s privacy or safety at risk. Much of the content on pedophile image-sharing websites originated as innocuous photos posted on social media (Battersby, 2015). Thus, sharing any child’s PII, including a picture that shows the face, is unwise as it compromises online privacy and potentially safety. Community policing pictures could still be posted, but children’s faces could be blurred or blocked with an emoji.

The goal of community policing is for LEAs to forge a stronger bond with their communities. Pasco County has done that in part by creating a viral daily social media event with a reach far beyond its county. In its #9PMRoutine social media tactic, @PascoSheriff is only asking participants to identify that their state is locked up with their event hashtag coupled with its ‘color the map’ gamification of the trend. However, the public nature of Twitter brings about issues that could come into conflict with the very idea of the #9PMRoutine: keeping citizens and their possessions secure. In addition to posting the request stated, nearly one-third of our sample also posted their city and/or county. While Pasco County does not ask for nor proactively collect children’s data, this study found that parents or other adults are sharing that information as part of their participation. Comparable to Fox and Hoy’s (2019) findings, individuals who have a public account are allowing anyone access to the information they share as they participate in #9PMRoutine. Cumulatively, all this information is much more identifiable, and when coupled with the users’ names being present on their Twitter profile, may lead to privacy and safety concerns.

Furthermore, disclosures of health-related information bring about a gray area in terms of privacy. While our study found that users are willingly disclosing the information, they are doing so to a safe and reputable source in Pasco County However, their privacy settings allow anyone with access to a Twitter profile, or Twitter’s API, to view that disclosure

Guidance for law enforcement agencies social media usage for community policing

The #9PMRoutine intent is to encourage development of habits related to safety and security (Daniels, 2019). However, it is possible that some members of the #9PMRoutine community could be jeopardizing their safety and security by the information they post in contrast to the tactic’s mission. How can Pasco County and other law enforcement agencies help protect their communities in these potentially unrecognized privacy and safety concerns?

Although our data did not reveal that Pasco County retweeted any of their participants’ posts to followers, all LEAs should be cautious in retweeting any content that contains citizen PII. Twitter injects tweets into users’ feeds that are liked by those they follow. Therefore, the Twitter platform is populating the agency’s thousands of followers’ feeds with these tweets, most of which contain some form of adult or child PII.

The International Association of Chiefs of Police’s (n.d.) current Twitter guidance does not mention privacy. LEA social media managers could be educated on these privacy considerations. Then LEAs could use their social media platforms to inform their communities of how to protect themselves and their privacy on social media. Tutorials on privacy setting restriction, awareness of who can see posts informing that the homeowner is out of town, children’s photos, or a variety of other personal disclosures could help protect community members. Promoting the protection of identifiable information such as city/county or neighborhood events could also ensure the safety of community members. Within #9PMRoutine, participants could be encouraged to use popular Internet gifs or memes that reflect locking up rather than posting anything personal.

Efforts to inform community members of how to protect themselves online will continue to add value to law enforcement agencies as online sources of information, while also contributing to their community policing mission. The ability to use their platform as trusted social media influencers to promote acts of safety in all aspects of life, not just physical security areas, could allow LEAs to continue building on the trust within their communities.

Limitations & future research

The study prompts many opportunities for future research. First, as noted in the method section, the Social Studio platform only provides access to Twitter accounts that are set to public. Whether by default or choice, these individuals are not aggressively guarding their online privacy. It is possible that other Twitter users who had private accounts participated in #9PMRoutine. An analysis of the PII they share and the engagement by non-Pasco Sheriff accounts would provide a deeper exploration into the potential privacy and safety concerns this engagement tactic is prompting.

This research focused on Pasco County as a case study, but we realize that other LEAs within the U.S. and around the world are also posting other forms of community-related content. Future research should examine whether this study’s results can be found across other LEAs that participate in this engagement tactic or offer their own, especially within the context of the country’s online privacy regulations (e.g. the General Data Protection Regulation—GDPR).

Future work could also look at individual Law Enforcement Officer’s (LEOs) professional accounts and see how they compare to the official agency account. Specific focus could be given to School Resource Officers (SROs) whose position has them daily engaging with children and often sharing pictures of the engagement. Further, LEOs, especially those who are canine handlers, are often invited to other organizations that serve children. Our study only examined Twitter. This study could extend to examine agency and officer Instagram accounts, which by their very nature require pictures.

As a method, content analysis can only describe communication. It cannot infer the thoughts, attitudes, or motives of the individuals who created the content. Thus future research could interviews or survey citizens who engage with LEA accounts in general and post PII to them in specific. Huang and Wu (2018) found that emotion enhances engagement and sharing of social media information, as reflected in the sharing of network-related information. Their observation relates to this study’s findings of private information freely being shared using the #9PMRoutine. Understanding if it is emotion, lack of knowledge of privacy guidelines on social media, or some other factor, such as a sense of community among the #9PMRoutine users, leading to these public disclosures can be integral in the understanding of online community behavior.

As the GSA (2014) notes, PII ‘requires a case-by-case assessment of the specific risk that an individual can be identified’. We have identified a previously unconsidered, and unintended, scenario that may put individuals at risk. Further, we have offered LEAs some practical suggestions regarding how they could parlay the relationship they already have with their community and extend the #9PMRoutine to also include locking up your online privacy.

Footnotes

Authors' note

Betsy Byrne DeSimone is now affiliated with Tennessee Tech University, Department of Communication.

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) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Alexander E Carter

Mariea Hoy

References

Aiello

(2018) Policing through social networking: testing the linkage between digital and physical police practices. The Police Journal 91(1): 89–101.

Battersby

(2015, September) Millions of social media photos found on child exploitation sharing sites. Available at: smh.com.au/national/millions-of-social-media-photos-found-on-child-exploitation-sharing-sites-20150929-gjxe55.html(accessedDecember2020).

Brainard

Edlins

(2015) Top 10 US municipal police departments and their social media usage. The American Review of Public Administration 45(6): 728–745.

Brainard

(2016) Fostering engagement through social media? The case of the Washington, DC, Metropolitan Police Department. In: Zavattaro

Bryer

(eds) Social Media for Government: Theory and Practice, 1st ed. New York: Routledge, Taylor & Francis Group, pp. 117–134.

Chanin

Espinosa

(2016) Examining the determinants of police department transparency: the view of police executives. Criminal Justice Policy Review 27(5): 498–519.

Daniels

(2019) Telephone interview.

Federal Trade Commission (1998) Privacy online: a report to Congress. Available at: http://www.ftc.gov/reports/privacy3/priv23a.pdf (accessed December 2020).

Federal Trade Commission (2000) Final report of the Federal Trade Commission Advisory Committee on Online Access and Security. Available at: http://www.ftc.gov/acoas/papers/finalreport.htm (accessed December 2020).

Federal Trade Commission (2013) Children’s Online Privacy Protection Rule; Final Rule. Federal Register 78(12): 16 CFR Part 312. Available at: https://www.ftc.gov/system/files/2012-31341.pdf (accessed December 2020).

10.

Federal Trade Commission (2015) Complying with COPPA: frequently asked questions. A guide for business and parents and small entity compliance guide. Available at: https://www.ftc.gov/tips-advice/business-center/guidance/complying-coppa-frequently-asked-questions (accessed December 2020).

11.

Federal Trade Commission (2018) Privacy & data security, update: 2018. Available at: https://www.ftc.gov/system/files/documents/reports/privacy-data-security-update-2018/2018-privacy-data-security-report-508.pdf (accessed December 2020).

12.

Fox

Hoy

(2019) Smart devices, smart decisions? Implications of parents’ sharenting for children’s online privacy: an investigation of mothers. Journal of Public Policy & Marketing 38(4): 403–413.

13.

General Services Administration (GSA) (2014) Rules and Policies—Protecting PII—Privacy Act. Available at: https://www.gsa.gov/reference/gsa-privacy-program/rules-and-policies-protecting-pii-privacy-act (accessed December 2020).

14.

Get Safe (n.d.) How burglars use social media to target homes. Available at: https://getsafe.com/how-burglars-use-social-media/ (accessed December 2020).

15.

Hanson

(2011, December 2) How social media is changing law enforcement. Available at: http://www.govtech.com/public-safety/How-Social-Media-Is-Changing-Law-Enforcement.html (accessed 18 March 2019).

16.

Hoy

Phelps

(2003) Consumer privacy and security protection on church web sites: reasons for concern. Journal of Public Policy & Marketing 22(1): 58–70.

17.

Hoy

Phelps

(2009) Online privacy and security practices of the 100 largest US nonprofit organizations. International Journal of Nonprofit and Voluntary Sector Marketing 14(1): 71–82.

18.

Huang

(2018) Understanding interactions between municipal police departments and the public on Twitter. In International conference on information, 25–28 March, pp. 37–46. Cham, Switzerland: Springer.

19.

Huang

Hou

(2017) Examining Twitter mentions between police agencies and public users through the lens of stakeholder theory. In Proceedings of the 18th annual international conference on Digital Government Research, Staten Island, NY, USA, June 2017, pp. 30–38. ACM Digital Library.

20.

IACP Center for Social Media (2010) Available at: http://www.iacpsocialmedia.org/wp-content/uploads/2017/01/Survey-Results-Document (accessed December 2020).

21.

IACP Center for Social Media (2015) Available at: http://www.iacpsocialmedia.org/wp-content/uploads/2017/01/FULL-2015-Social-Media-Survey-Results.compressed.pdf (accessed December 2020).

22.

IACP Center for Social Media (n.d.) Twitter fact sheet. Available at: https://www.theiacp.org/resources/technology-and-social-media (accessed December 2020).

23.

Kerlinger

(1986) Foundations of Behavioral Research, 3rd ed. New York, NY: Hold, Rinehart & Winston.

24.

Lortz

(2016, December 28) What is community policing? Available at: https://www.everbridge.com/blog/what-is-community-policing/ (accessed December 2020).

25.

Meijer

Thaens

(2013) Social media strategies: understanding the differences between North American police departments. Government Information Quarterly 30(4): 343–350.

26.

Quinn

Grove

(2018) ‘The whole window just shatters into a million pieces and it doesn’t make any sound’: an examination of ‘affordance’ and situational cues for theft from a motor vehicle. Security Journal 31(4): 841–858.

27.

Schneider

(2017) Body worn cameras and police image work: news media coverage of the rialto police department’s body worn camera experiment. Crime, Media, Culture 14(3): 449–466.

28.

Walsh

O’Connor

(2019) Social media and policing: a review of recent research. Sociology Compass 13(1): e12648.

29.

Williams

Fedorowicz

Kavanaugh

, et al. (2018) Leveraging social media to achieve a community policing agenda. Government Information Quarterly 35(2): 210–222.

30.

Wimmer

Dominick

(1997) Mass Media Research: An Introduction. Belmont CA: Wadsworth Publishing Company.

31.

Yin

Fabbri

Rosenbloom

, et al. (2015) A scalable framework to detect personal health mentions on Twitter. Journal of Medical Internet Research 17(6): e138.

32.

Zawisza

Garza

(2017) Using an eye tracking device to assess vulnerabilities to burglary. Journal of Police and Criminal Psychology 32(3): 203–213.