Police communications and social media

Abstract

This article considers the affordances of Web 2.0 social media tools for public communications by the police and illustrates their use, presenting data from exploratory fieldwork with officers and staff in five UK police forces. Future lines of development are noted that may merit ongoing research.

Introduction

Police forces engage in multiple forms of communication, for numerous purposes, with a variety of recipients and stakeholders. This study focuses on public-oriented communications but recognizes their internal dimension. Information is a malleable entity that changes as it flows through forces to Communications Departments (‘Comms’) and outwards. The communications in frame may relate to specific investigations, focused information campaigns, or broader objectives (for example, reassurance). Issues of provenance and trustworthiness play differently in each. In a ‘post-truth’ era of rising public scepticism, ‘fake news’, trolling, cyber-bullying, and cybercrime, one might speculate whether the public expects more of police online communications than of communications from other sources.

Whatever their prime purpose, police communications often have a supplementary aim, namely, relationship-building. Police–public communications long cleaved to instrumental purposes while neglecting the influence of social media (‘SM’) on the police–public relationship. Some feel that, by doing so, forces use SM as simply another broadcast channel rather than for its core function – the social. Is relationship-building what the public expect? When a force adopts that approach, does it undermine legitimacy?

Research design and methods

This exploratory study drew on research literature; practice examples from group discussions in an academic/police collaboration around counter-terrorism and serious crime; and key informant interviews. Relevant literature included studies of online information flows and evaluations of police SM communications. The practice examples concerned SM-based interventions and campaigns. The key informant interviews took place in 2017/18 and drew on a non-probability sample of officers and civilian staff having the communications function in five police forces with a combined population of 8,089,200. Their communications personnel numbered respectively 15, 27, 25, 14, 19 (average = 20). Since key informants are individuals whose role gives them extensive insight into the topic of interest, the technique is associated with ‘small N’ exploratory studies (Payne and Payne, 2004). Five interviews were carried out per force, using a selection matrix of roles and whether the role incumbent was headquarters-based or operational. Interviews pursued instances of good/poor practice perceived by respondents; views about the role of SM in force communications strategies; appraisals of obstacles to SM’s potential; assessments of what communications tasks were unsuitable for SM.

The key informant interviews were semi-structured. Exploratory data analysis using MAXQDA qualitative software supported a code-and-retrieve inspection of data from which analytic themes were derived and refined. Data extracts are presented from five informants, each from a different force, whose responses were particularly pertinent to the themes pursued here and who had operational policing experience: the Senior Communications Manager of a shire force with two large towns, the Digital Communications Manager of a suburban/rural force, the Digital Communications Manager of a large urban/rural force, the Social Media Manager of a second large urban/rural force, and the Head of Marketing and Internal Communications of an urban/rural force. Where their comments reflect those across the overall sample, this is noted. All had several years or more experience in communications work. SM featured in their off-duty lives, for identity work as well as for communication (Suler, 2015). All mainstream SM platforms were used across the sample bar Flickr. Twitter and Facebook were the principal tools of police SM communications, with specific purposes for using particular platforms often cited (including the kind of social network they tap into) and others having niche utility. This discussion is thematically driven rather than presenting points platform by platform.

Digital affordances

‘Affordances’ (Gibson, 1966) is a fundamental concept employed to understand how new technologies are used. By examining benefits and limitations it can provide new insights. For instance, applied to Web 1.0, it revealed that the familiar process by which new devices are configured by users to meet their requirements is paralleled by a process in which new devices ‘configure the user’ (Woolgar, 1990). The user’s behavioural repertoire is changed by the techniques they must follow to use the features provided by the technology. Adopting a new technology can change how users seek to solve problems for which they see the technology as a solution (Aiken, 2016). The policing corollary of the ‘configuring the user’ phenomenon is the extent to which policing has adapted its behaviours to SM’s affordances.

In learning a technology, users also discover new needs that it can satisfy. Sometimes what its developers foresaw as the technology’s primary use is not its primary use in actual practice. For example, the telephone was originally conceived as a broadcast technology like radio (Bijker, 1995). Rather than peer-to-peer communication, it was originally expected that the switchboard would broadcast programmes to the mass of individuals dialling in.

Thus, a technology’s affordances are not only those for which it was designed but those that users discover it can support – sometimes at the cost of efficiency, effectiveness or ethics. In open plan offices it is more efficient to visit a colleague and speak to them, but often an email or text is sent instead. E-mail and text are not as efficient but their hidden affordance is that one can simultaneously have separate ‘conversations’ with several people. Further, email or text is retrievable in a way that speech is not. Like other IT, mail/text are self-documenting, providing a further dimension to communication. This may or may not be beneficial. If the communication concerns an important work-related matter, then a retrievable record is a boon. If it is an insult, one or both parties may subsequently regret there being a retrievable record. Such considerations feature in deciding a given technology’s effectiveness for given purposes. In the SM context, this raises the point that not being self-documenting may be a desirable affordance in certain communications; whereas Facebook is persistent, Snapchat is not.

The affordances concept tells us that (i) technologies change how their users behave; (ii) users actively work to identify new needs that are met by the technology; (iii) the spell of technology is sufficiently strong that it can displace previous approaches that were more efficient, effective and/or ethical.

Socio-technological studies employ an ‘adoption curve’ to measure uptake rates of new technologies. Certain population segments are predisposed to be ‘early adopters’, with the majority following later, and a minority roughly equivalent to the early adopters remain disinclined to adopt (Fielding and Lee, 1998). Typically, early adopters are young, well educated and relatively affluent. However, circumstances make a difference. Limited broadband and technical infrastructure made for a different profile of early adopters of mobile phones in sub-Saharan Africa. Mobile phones provided access to the Web not only for young Africans but for a wider age spectrum.

Web 2.0 was explicitly designed with SM applications in mind. Web 2.0 is characterized by websites that support user-generated content, are interoperable (with other products, systems, and devices), and are easy to use. Social media are interactive Web 2.0 applications. Users create a profile on the application and the SM service connects it with that of other users, enabling the creation, development, and maintenance of social networks. In the apt words of an anonymous Wikipedia entry, SM created ‘substantial and pervasive changes to communication between businesses, organizations, communities and individuals’.

Digital affordances and police communications with the public

Policing’s engagement with technology is often relatively isolated from technological engagement in other social institutions. Concepts developed around technological engagement elsewhere may not apply, or may apply differently. Although academe’s engagement with police SM use is quite recent, distinctive conceptualizations are emerging. An instance is the ‘cyber-neighbourhood’ concept (Williams et al., 2013), which adapts the rationale of community policing to the SM arena, for example, in considering the role of central figures within informal social networks who can pass useable community intelligence to police based on numerous everyday contacts that yield granular information. Originally understood as an offline function performed by persons such as hairdressers, delivery drivers, and concierges, this has its equivalent in the SM arena’s ‘influencers’ and ‘followers’, but with the difference that social networks can be systematically identified and communication flows analysed. Williams et al. (2013) demonstrate that such information can be overlaid with official statistics (crime, social deprivation) to compare ‘terrestrial’ and ‘cyber’ streets, supporting a practice of ‘neighbourhood informatics’.

A considerable spur to criminology’s engagement with SM in policing was the 2011 riots that spread from London to other English conurbations and even small towns. This not only provided a testbed for applying ‘Big Data’ techniques to social problems but ensured the research received media coverage. Prime Minister Cameron and Home Secretary Theresa May, amongst other Conservative politicians, had castigated the rioters as ‘feral’ sub-humans motivated purely by greed (not political motivation or need due to deprivation), also claiming the rioting was organized via SM, as did senior police. This claim disintegrated when confronted by sophisticated statistical analysis of a massive volume of the SM traffic during the rioting. The predominant use of SM had not been to organize rioting but to organize the spontaneous clear-up parties that put out fires, boarded broken windows, and swept the streets (Procter et al., 2013). This benign use of SM contrasted with the grossly inadequate calibre of police communications during the rioting, including the complete failure of an over-budget bespoke police radio system during the worst rioting. As to the greed explanation, careful analysis by the London School of Economics of individuals involved in rioting and who appeared at court found that they occupied the classic social deprivation profile of poor schooling, low educational attainment, and poverty (Newburn, 2012). Sentencing had nevertheless been severe, with courts hastily hearing cases within hours of Cameron’s pronouncements. Importantly, the dismal police performance in the riots was a signal moment in their engagement with SM. Subsequent public disorder has seen increasingly effective exploitation of SM by the police, including cultivating citizen ‘influencers’ to counteract false information in real time during rioting (King, 2013). Case studies of techniques exploiting SM affordances in strategic but responsive ways appear in Innes et al. (2018).

Outside dramatic disorder, the police have long sought to exploit successive mediums of communication to engage the public; prominent senior officers have depicted this as a major element of public accountability, with a role in securing public support and police legitimacy. Manning’s (2003) semiotic study of processes of encoding/recoding employed in everyday police–public communications demonstrates similarities in communicative tropes and narratives between police agencies on both sides of the Atlantic, and his study of the UK inspectorate responsible for maintaining safety at nuclear installations (Manning, 2008) demonstrates similarities to those of other agencies that seek to reassure the public that they are ‘in control’. However, much of our conceptual apparatus for police ‘image work’ (including the term itself) originated with Mawby’s (2002) seminal study of police press and public relations officers. Mawby links the fact that policing is of keen interest to the public – forming a substantial topic of much entertainment and documentary reportage – to police themselves seeking to influence, if not steer, their image as projected by the mass media. That agenda endures but is delivered in new ways, with SM enabling a greater degree of interaction between police and target audiences.

It is important to avoid hyperbole, and there is evidence (discussed later) that the forms of SM interaction described by respondents to the present study, although significant, are not yet predominant. Indeed, Bullock (2018) argues that the police organization and its occupational culture remain substantial obstacles that have to date eviscerated the trumpeted ‘transformative’ potential of police SM activity. The advance of digital technologies has modified aspects of the role of police communications officers, but much remains the same. The tension between contemporary affordances and established constructions of the role informs the interview data reported here. Mawby wrote of police media output as ‘promoting, projecting and protecting’ the police image, and that alliterative summary still pertains, but under a different balance of McLuhan’s ‘medium’ and ‘message’. As to legitimacy, Mawby makes the important point that image work can both promote it and mask problems in securing it.

Mawby (2010) subsequently revisited Chibnall’s classic study of the relationship between police and journalists, arguing that the traditional police press office and crime reporter have been superseded by a ‘corporate communications’ mode that is increasingly asymmetrical in favour of the police. The simultaneously symbiotic and frictional relationship between reporter and press officer that Mawby (2010) discusses has, to an extent, moved within forces, where command and middle rank officers adhering to a corporate communications model confront the opening of new channels brought by Web 2 technologies. In that regard, it is not only the police who exploit SM to steer how events are understood.

Innes, originator of the ‘signal crime’ concept (Innes, 2014), has latterly examined SM commentaries following terrorist events (Innes, 2020), documenting techniques used to propagate disinformation. He develops a conceptualization analogous to that of Sykes and Matza’s ‘techniques of neutralization’ (1957), but where it is information that is deviant, not behaviour. Response to ‘fake news’ was a preoccupation of our respondents. Although the examples reported here were more mundane than those Innes cites (with three exceptions – a major disaster, a terrorist plot, a homicide), the respondents had to deal with neutralizations such as denial of the victim.

This section has profiled a mainstream policing mode facilitated by police use of SM analytics (cyber-neighbourhood policing); a major public disorder event where police failed to make adequate use of SM (2011 riots); and foundational sources on police presentation strategies (so interested readers can pursue the burgeoning field of police/public communications).

Evaluating SM in police communications involves comparing different modes of communication: SM versus ‘conventional’ modes such as newsletters, static Internet pages, newspapers, radio and TV; and SM versus ‘traditional’ modes such as public notices or officers knocking on doors. When comparing SM with conventional modes there are trade-offs between trustworthiness and reach, interest and usefulness (Ruths and Pfeffer, 2014). Mode comparisons should bear in mind the different purposes for which police forces communicate with the public. These include Publicize (for example, a successful police initiative); Advise (for example, how to source and install crime prevention hardware); Inform (for example, how to recognize the signs of domestic abuse); Warn (for example, a terrorist alert); and Appeal (for example, requesting public help in seeking a criminal suspect or missing person). Underlying these is the requirement to Engage. This is often neglected against more immediate and/or instrumental purposes. SM has potential for building public engagement, and the quality of everyday police–public encounters is a prime contributor to public confidence (Police Foundation, 2015).

Empirical material

Readers are reminded that this is an exploratory study; resulting data should be treated with caution. It points towards future lines of research rather than being definitive. Empirical material is presented relating to the six communications ‘purposes’ noted above, while recognizing overlap between them. The purposes comprise different inflections rather than being wholly distinct and self-contained.

Publicize

Respondents identified the main instances of the publicize purpose as disseminating the detection of wanted and missing persons and successful prosecutions, where ‘social media can play a massive part’ (DI [respondent ID]: 3 [transcript page]). Initially, police posts simply publicized press releases and a link to the force website:

They weren’t using social media in the way the rest of the world uses it . . . So we turned off automatic tweeting and the automatic tweet going to Facebook and got the team to start writing posts for the platform they were using [and] in a much more friendly and human way . . . Now whenever we publicize a good initiative we think about the platform we’re using and write something that works in [it]. (KB: 3)

A common technique involves identifying popular events for which SM can demonstrate a policing dimension. For example, on Valentine’s Day ‘we put out messages to our “long lost loves” . . . five Wanted Persons . . . [I]t got a lot of attention’ (KB: 3). This led to an arrest, ‘so we publicized the fact . . . and thanked the public for their help’ (KB: 3). Novelty registers with the public: ‘We’ll publicize if they’re doing something new to combat drugs or dangerous driving or off-road bikes . . . People listen more if they can see we’re looking at new ways of trying to tackle a particular crime’ (IG: 2). Although journalists still sometimes accompany officers on operations, ‘our main success is social media . . . So we’ve done Facebook Live [on] “life with a response officer” . . . showing what [they] have to deal with on a standard night in central [very large city]’ (IG: 2).

Forensics teams and dog units also receive such coverage. Instagram Stories and Snapchat are used to reach younger people (IG: 2). Photography and visuals teams exploit SM video functionality not only for dramatic material (‘knocking down doors, drugs operations and raids’ – KT: 2–3) but for crime prevention messaging and encouraging crime reporting.

Advise

There was consensus that SM content about crime prevention and advice like safe passwords and data security was worthy but dull (DI: 3–4). ‘Crime prevention is one thing they’re most disinterested in on social media . . . Only 11% of people were interested [citing a Comms team survey]’ (EC: 3). There was also consensus on how to arouse interest. ‘People are most interested in updates about incidents happening now in their area . . . For normal low-level crime prevention advice [we] tie it up to incidents. We get far more reaction if there’s something that makes it relevant’ (EC: 3). Advice on locking windows at night would be targeted to an area where a residential burglary had recently occurred with entry gained through an open window. Traffic units attract high Twitter followings ‘and they will put out real cases of somebody they’ve just caught and provide advice . . . [If] they’ve caught somebody with bald tyres they will say, “It needs to be this depth”. . . It works better if it’s alongside a real story that’s just happened’ (KB: 3–4). An instance using Snapchat involved an individual advised for leaving his car unlocked:

Then they found drugs in that car, gave him a caution but [also asked] ‘Do you want to talk to our younger audience about drugs and your life?’ because they’d messed up his life, and he did, which was a great bit of advice and it was real and it wasn’t ‘Hey, lock your doors’. (KB: 3–4)

Targeting advice requires gauging where specific groups start from in receiving such messaging:

We have done a cybercrime online campaign about preventing rape and serious sexual assault and building confidence to report [them] . . . We did benchmark testing on perceptions of the police . . . [before and] afterwards [asking] ‘did the campaign move your opinion positively or negatively?’. That was targeted to age profile. (EC: 3)

Advice is ‘a subject where we can come across very police-y, telling people what to do’ (KB: 3–4), so style and tone are configured to target-appropriate platforms. Blanket messages could pose dilemmas if recipients saw them as opportunities to respond. ‘We will provide generic advice, but . . . we can’t take reports of crime. So it’s signposting to [other] organizations, or “thanks for letting us know”’ (KT: 4).

Facebook and Twitter functionality is exploited in interactive videos and ‘wideos’, cartoon-like short videos to reach specific audiences. Another technique is coordinated messaging. In a burglary campaign, Comms aligned all Twitter account headers and Facebook profile pictures ‘so it’s all pulling together’ (IG: 3). Scale is substantial; one large force managed over 260 Twitter accounts, also disseminating campaigns via an email database. Impact remained uncertain. ‘It’s very complex to measure [the impact of] a crime prevention message based on social media . . . You can [only] measure the level of engagement you get with that post or clicks through to a website’ (EC: 3).

Inform

Inform is a wide-ranging purpose that extends from reporting major road accidents to recognizing grooming or domestic abuse. A four-week domestic abuse campaign at a time of peak abuse statistics took over all a force’s SM channels. Themed images coordinated banners and profile pictures. Hashtags using memorable phrases from the victim’s perspective appeared in each tweet. A flash mob rally at a mainline station displayed placards with the phrases, and Snapchat posts worded the phrases from a youth perspective.

Another sensitive campaign targeted the family and friends of hardcore pornography users. Some messages encouraged users to contact police if wanting help, while others targeted family members/friends, with wording advised by a charity. Collaboration is common, and required for official campaigns (KT: 4–5). One campaign funded by a Police and Crime Commissioner targeted drunkenness. A previous campaign had tried scare tactics – ‘one punch could kill’. The new campaign highlighted the negative effects on nights out. Before/after drunkenness arrest rates were compared to gauge impact.

Warn

Although some warnings are routine – such as warning of hazardous sea conditions at popular beaches – this purpose particularly concerns major events. Here Twitter is pre-eminent. It is seen as ‘a news channel . . . where people will go first’ (IG: 3). Some forces run Twitter accounts with the force name followed by ‘Breaking’, promoting their use for early warning. Twitter Alerts warn about imminent as well as ‘live’ events. Alerts are followed by detail on Facebook.

All UK primary accounts have Twitter Alert functionality; we used that last Sunday for a large waste depot fire to warn people. We also did immediate Facebook content including video to warn and inform about keeping windows and doors closed, keep away from the scene. (EC: 4)

Twitter warnings end with steers to Facebook.

SM also has latent functions. There is an association between disorder-related Twitter traffic and police-recorded disorder (Williams et al., 2017). Counter-terrorism police monitor SM networks for indicators of community tensions. Dedicated applications include COSMOS, developed by academics, but UK forces commonly use COSAIN, which provides sentiment extraction and filtering rules informing predictive analytics that identify relevant content. COSAIN’s developers emphasize design input from law enforcement personnel; its support for user-defined risk/threat levels; its capacity for real-time filtering; and that users can build pre-defined trigger points. However, this functionality is not unique, nor is the range of Web 2 entities it searches. COSMOS’s ‘tension engine’ provides similar functionality, was designed in consultation with police, and captures additional socio-demographic data sources. Programs with a commercial rather than academic base seem attractive because of perceived assurance of continuity of support. Representing the police voice in the ‘cyber-neighbourhood’ led respondents to value reliability because quick response was essential. Speed is critical when warning:

If something really big is happening you have to do something . . . quickly because . . . you want the police to be the credible trusted voice [from] the start. We’ve had situations [where] we don’t know what’s going on, ‘we’d better wait and write a press release’, by that point an hour has gone, people are following other accounts or the media . . . Get out there with something quickly, even just a holding line to say ‘we’re aware, more info to follow’. (DI: 5)

At a major incident with double-digit fatalities ‘the people at [site] first was the Media Relations Team [who] didn’t get to social media quick enough, we had . . . the public asking us what was going on but we hadn’t replied [or] even put out a statement’ (KB: 5). Once on SM, targeted posts were used, with some warning spectators to stay away and others warning people nearby to stay put so they could be led to safety. There were regular SM updates for a week and the force launched a new Facebook page for the town where the disaster happened to provide localized communication during the recovery stage.

It was important to know one’s audience, for example anticipating that the national ‘Run, Hide and Tell’ (counter-terrorism) campaign was less relevant outside major cities (KT: 6). A major event required several forces and an armed police presence for 10 days. Joint Operations Command created a dedicated Twitter feed, anticipating public resentment.

We’d had a lot of criticism beforehand and . . . people protesting. [But] it really turned round . . . with press officers providing positive imaging of what the officers were there to do . . . [Non-residents] used it as traffic advice, places to avoid, and information of what was going on. (KT: 6–7)

Multiple purposes can be pursued once a channel has secured engagement.

Appeal

SM was valued for appeals for help with missing persons. A dramatic example involved ‘using Facebook at 2 a.m. on a Saturday . . . It was only through a Facebook audience that we’d already built that we reached the person with the information that led us to find the woman unconscious in her car, overdosed. We had no [other] communications channel but were able to target it to the area where we believed she was’ (EC: 5). As an online presence, forces could tap into knowledge of immediate value. A missing child or vulnerable person’s ‘picture goes out very quickly and you can very quickly get people commenting on where they are’ (KT: 7–8). Internal analysis found ‘nearly 60% of appeals resulted in some kind of information and 95% of the information . . . was a result of SM. Something from radio, very little from local papers’ (DI: 6).

Building on an initial incident or event characterizes wider SM strategies. Describing the Valentine’s Day campaign noted earlier as ‘culture bombing’, a respondent observed: ‘we’ll jump on something that we know they want to talk about . . . because people don’t expect the police to do this.’ Extending their reach was as valuable as an arrest:

[W]e want to build relationships before asking the public to do something . . . The algorithms in Facebook and Twitter favour pages that the audience actually engage with – I might have ‘liked’ [force] but if I’m not liking or commenting on their posts, Facebook will think, ‘they’ve liked the page but actually they probably don’t want to see their content’. (KB: 5–6)

Exploiting ‘fun stuff’ increased the audience for appeals.

We did the Running Man Challenge [an Internet fad] . . . That reached over a million people, engaged about 60,000 people and we got loads of likes. Two days after that we had an elderly missing person that we appealed for [on Facebook] and that post reached more people than any of our misspers previously. (KB: 5–6)

Other platforms had their niche. Flickr albums enable uploading locale-specific CCTV. A force kept its ‘Most Wanted’ gallery on a Flickr page, along with CCTV appeals also posted to YouTube. Both formats referred viewers to Facebook for details (IG: 3–4).

Engage

If there is a mainstream in police SM work it is engagement: ‘a strand . . . across all five [purposes]. Engagement achieves the results in all those other five’ (EC: 5), a view shared by all respondents. Engagement involved getting communities into a discussion ‘and building relationships with them’ (KB: 3). This underlay choice of platforms, Facebook being the ‘workhorse’ owing to its reach and support for extensive content whereas Twitter was ‘more broadcast due to the sheer volume’ (KB: 3). Online and offline could interact, for example, with Twitter used to draw users to web pages publicizing force open days, then encouraging them to join the planning committee.

Although engagement has indirect benefits (for example public satisfaction ratings), direct payoffs involve collaboration.

The riots in 2011 really kicked off our SM footprint. For us to stop rumours . . . we got on board with a local blogger; he was getting more followers each day and we . . . managed to build our own profile up because the way he was wording his messages people knew he was getting information direct from us . . . He was also directing people to our SM, so people could see he was trusting us . . . we were getting his followers to follow us to get the latest updates and we’ve managed to increase from there. (IG: 4–5)

Engagement requires accommodating local interests:

We’ve steered away from the Manikin Challenge, you’ve got to know your community and people in [force area] wouldn’t really appreciate it . . . They created a personality for an officer in London, ‘Dancing Dan’, which is great, but what if Dancing Dan arrests you? You’re not dancing now . . . But football and rugby, one of the biggest hits was when we had the haka, an officer doing it off his own bat, the engagement . . . was incredible.’ (KT: 9–10)

Spontaneous acts made valuable copy, such as officers buying presents for a family burgled at Christmas, requiring only plain reporting – ‘their natural instinct’s creating [the story]’ (KT: 9–10).

The impact of SM communications

Comms officers agonized about impact:

Social media is great for awareness-raising, but how do you measure whether that has actually led to anything? As a result of that post someone has changed all their passwords and made them more secure, you can never find out . . . unless someone says ‘Great advice thanks, I have gone away and done that’ . . . It’s very few and far between . . . Re-tweets, shares, views, you could definitely say people are more aware of the content. But whether that has led to ‘I’ve gone and bought a padlock, I’ve changed my password, I’m not driving as fast’, it’s mainly impossible [to know]. (DI: 4)

Nevertheless, some techniques helped, such as combining online surveys with embedding different captured links in different channels ‘so we can identify somebody who’s running from a Twitter page to a Facebook page etc’ (EC: 6). It was sometimes possible to ask directly – Open Day participants were asked where they heard about it (70 percent said Facebook or Twitter; EC: 3).

Forces use Google Analytics, though some only on their main website. Some forces use SocialSignIn, a suite of programs including a sentiment analysis tool. Disambiguation was a substantial problem with sentiment analysis. ‘“Arrest” is a negative word in terms of sentiment, but in policing terms good news’ (DB: 6). Time ‘manually changing sentiment and really looking into it’ was a disincentive. ‘There isn’t a tool out there that works for policing . . . Because you see a headline “Murderer put away for 20 years thanks to great police work”, that [would] come up as negative because of the word “murder”’ (DI: 8). Nevertheless, workarounds exist. SocialSignIn reports who is messaging about force communications and scores how influential these ‘top engagers’ are. Influence can be combined with how much the user ‘likes’ force communications. ‘We can look at all their previous interactions with us; in Twitter at what hashtags they use, other people they talk to and the sentiment overall of that user’ (KB: 7). Previous interactions fashion how Comms teams or force contact centres respond to an individual, using tools such as Klout (more recently, BuzzSumo).

SM can subtly make Comms teams the voice of an alternative perspective within forces:

Our corporate main account [pages are] full of bad people wanted for recall to prison. It just made people more feared of crime . . . You would see the image and think, ‘that’s really frightening, it must be that all of [force area] is like this’. It just wasn’t representative of the image we’re trying to do. (KT: 6–7)

Although police justifiably want to catch people who breach prison recall, the respondent developed the point around targeted messaging:

We worked a lot in terms of not being the kind of force that you have all the images and documentaries of knocking doors down, fast cars, blues and twos, because a lot of our work is about . . . working with communities and low key stuff . . . We try to weed out fear factor . . . If we were looking force-wide then a Wanted would be appropriate, but if it was only in [large town] then why did people [elsewhere] need to know. (KT: 7–8)

Such thinking informed a drive for creating more localized accounts.

Tailoring social media communications

Communications challenges include reaching a particular demographic or locale, and making contact when time is critical or coverage is an issue. Forces hold privileged information that facilitates targeting and tailoring, such as victim and offender profiles, but sensitivities make nuance important. ‘There’s a real danger [with a rape awareness campaign] . . . of saying “if you [have this [profile] . . . you are vulnerable”, [it hints at] victim blaming’. To make targeting discreet, ‘we used [profile information] at the back of the systems to make this post visible to females between 25 and 34 who live in this area and mention these things on Facebook so it wasn’t an overt use of that data’ (EC: 3–4). Similarly, the vehicle licensing database assisted a child sexual exploitation campaign geared towards spotting signs.

We have very specific audiences for that – professionals, taxi drivers, parents. All with slightly different messages and different platforms . . . For professionals we were getting to them through LinkedIn, and hotels we were sending emails through our neighbourhood alert system, and [for] parents, Facebook. (KB: 4–5)

Differentiation extended to tools within the tools:

We used Facebook to target two postcode areas where there was a rape . . . Instead of getting a couple of PCSOs [police community support officers] to deliver leaflets, which would have cost us £500 [plus] the man-hours, we spent £120 in Facebook to specifically target these postcodes with a little ad that said, ‘We need your help, talk to us’. (KB: 4–5)

Within two days 30 calls were received, yielding information leading to an arrest. Whole-town leafleting and broadcast SM messaging would have elicited hundreds of responses, requiring filtering, so ‘they wouldn’t have caught them as quickly’ (KB: 4–5).

Tailoring involved assessing the target audience’s social context.

We can reach people who aren’t online through their families if we tailor the message right. Targeting people in their 40s about telling their parents about bogus callers is better than [broadcast] messaging . . . [A] child ringing up their elderly parents will have more impact than them hearing that on a radio station. (EC: 7)

Tailoring existed before SM but was resource intensive. Such interventions were justified in counter-terrorist investigations. ‘Comms produced 500 DVDs to post through letterboxes in . . . a tight-knit community, terraced back-to-back housing in a Muslim community’ (EC: 4). The DVDs went to addresses near the suspect address and were used owing to low digital literacy amongst female residents and because DVDs elicit more attention than leaflets. A local commander spoke directly to camera, stressing that females who did not go out to work were especially likely to have relevant information. Information received resulted in ‘executive action’ by armed officers.

Tailored communications must accommodate critical or inflammatory messages, most often encountered in highly localized contexts.

Some of the worst in terms of negativity are local appeals, because it’s people that know who [the suspects] are. [T]hey’ll argue with each other, families, and they often go quite wrong very quickly, people starting to make threats and having a go at each other. (KT: 5–6)

Knowing when to intervene involves fine judgement. For example, a force Facebook received posts criticizing an intervention but this was allowed to run until others defended it (KT: 6).

Big operations excite interest but people mostly want ‘hyper-local’ information. SocialSignIn is used to analyse Facebook accounts and see whether residents are coming to the website through a local station’s account, a district account, or an adjacent force’s. Some forces assign specific Twitter and Facebook accounts to the most populous areas of each of their districts. Others operate multiple accounts based on function (Response, CID, Traffic). In major incidents or high-profile cases, teams assume citizens will monitor the force’s main Twitter account and the relevant police station’s. Focus was on the main account but, because it was less granular, posts built up the human interest aspect, such as the ‘personality’ of a town or squad (KB: 7–8). Some forces treating Twitter as a breaking-news platform extend it with neighbourhood alerts. Users subscribe to areas such as their neighbourhood, workplace, and/or children’s school area, receiving crime alerts, trends, crime prevention images, and area events, which are filterable (IG: 6).

Engendering trust

Trust is a major issue for SM communications (Suler, 2015). Respondents were generally sanguine, believing that forces enjoy high trust, but garnered it with responsive, interactive communication exploiting mode-specific affordances.

Every platform and medium, from traditional to digital has certain trust issues depending on who is communicating and the receiver’s world view . . . Twitter or Facebook now [have] only got verification on the main force account . . . We have people [who do] not believe that [city] police is actually the real [city] police because they don’t have verification. (KB: 9; policy current during fieldwork)

However, technical aspects were only modifiers of qualities marking any sound relationship. ‘If . . . people are being negative or asking questions you don’t shy away . . . because other people are watching those conversations’ (DI: 10). SM was not simply ‘a transmitting tool’ – that suggested propaganda. Appropriate ‘tone of voice’ avoided formality, pomposity, and advertising agency slickness:

Would you say to somebody on their doorstep, ‘the officer was progressing in a northerly direction’? You wouldn’t. Talk to people on social media like you would in person . . . You can still be professional but you have to be human. (DI: 10)

Spin was also an issue:

You have to put it out quite straight without a slant, because that’s showing your organizational [perspective]. [A]s a police force, we’re meant to be ‘this is what we do, this is how we’re going to stop it . . .’. It’s far more trusted. (KT: 11–12)

Clarity of purpose was best signalled by directness and candour. ‘Essentially we are a force, and people often don’t like us very much’ (KT: 11–12). The goal was not friendship but working relationships.

I’m fine with doing capacity-building pictures of police puppies . . . the dog bouncing around in the snow on [our] Facebook page [is] about building reach and capacity for the future, [but] nice soft PR is not what the public always want . . . If it’s about informing them and driv[ing] them to a wider debate you’re looking for lots of engagements on the conversation. [I]n the serious sexual assault [initiative] we deliberately provoked people to engage in a conversation about ‘can you give consent if you’re drunk?’ That way we’d reach more of the audience [and] get people to form an opinion. (EC: 7–8)

Opportunism also features:

At the time [of the 2013 floods] we only had about four thousand followers on Facebook. Within two days of flooding there was a Flooding Group that had nine thousand people. We had spent four years trying to create our Facebook following and this group had created it! So . . . we need to go where the conversation is happening. (DI: 7)

The respondent perceived three channels:

The people that are talking to us directly on social media, so they have @mentioned us or written on our wall. The second is people that haven’t @mentioned us but are talking about [police interventions] . . . on social media. We are pretty good at the first one, not bad on the middle, but the third, the big gap, is people talking about policing matters [on another channel]. How do you reach people talking about a burglary that has happened down their road in their community Facebook group? I’m a member of a number of local Facebook groups, they talk about policing issues and lots of rumours. All it would take would be for the police to come in on that and say ‘these are the facts’. The whole conversation would settle down, confidence in police would grow. (DI: 7)

Obstacles to the use of social media

Although organizational problems in police SM communications included technical legacies (for example, a year spent migrating to Office 365), the telling problems were human. Senior officers were often late, naive adopters of SM, impressed by numbers of followers and re-tweets without grasping the messages of network analysis regarding influence and targeting. Instances of senior naivete included national policy guidance drafted using a Wiki because command officers did not realize the security implications; one of Britain’s largest metropolitan forces not creating a Facebook presence until 2014; and ranking officers who thought forces could turn off Twitter if they didn’t like what was being tweeted. Some senior officers dared not admit they didn’t understand SM, others saw a panacea not requiring structured evaluation.

Social media is thought of . . . [as] a way to do something for free . . . ‘We’ll just do something on social media’ so you can tick the box. ‘Look, it’s National Stalking Day. We haven’t got any budget, it’s not a massive priority . . . To be seen to be doing something we’ll do Facebook and a tweet’. (DI: 9)

This led to engagement being seen as its own objective rather than a tool pursuing strategic objectives.

Respondents suggested ‘getting’ SM required an ‘epiphany moment’, such as a chief constable who saw colleagues using it to direct interventions in the 2011 riots and derive intelligence in real time (DI: 10–11). In contrast to those who shunned SM were senior officers who ‘we have to rein in because [their sends] are not appropriate’ (IG: 7). There was surreptitious Hootsuite monitoring of quite senior people and ‘sometimes we get caught on the back foot [by journalists] where an officer has tweeted something . . . they shouldn’t have’ (IG: 7). Chiefs can have ‘quite heated discussions on Twitter that we wouldn’t feel . . . appropriate for us, but when you’ve got an [assistant chief constable] talking . . . and he ends it in a certain way, you think, “Okay, he’s listened to that point of view, doesn’t agree but it’s kind of ended it”’ (KT: 8–9).

It was apparent that SM staff occupied a niche and perceived skillset that could challenge the division of labour. ‘People in Corporate Comms always come to me when they’re about to do something with social media . . . because they are not used to writing like that’ (KB: 9–10). SM was pigeonholed as ‘something for the Comms team’, potentially causing overload and mission drift.

Forces are using their Comms team to answer first contact-type messages . . . Somebody tweeting ‘Why is there a police car parked down my road’, how is that any different to somebody phoning and asking the same question? It’s not. So why is the Comms team answering questions like that? . . . [B]ecause it’s on that channel. (DI: 11)

Comms teams’ emergence from press release mode into a more direct relationship with the public can cause friction between operational imperatives and engagement:

If something’s happening and we’re slow to respond it’s already been played out on Twitter . . . We called [murder of a couple outside their workplace] an ‘incident’, people were tweeting ‘I can see a body, it’s not an incident, it’s a murder’. But we couldn’t confirm that because the family hadn’t been told . . . We also asked people to refrain from posting those pictures [but] you can’t control that. (KT: 12)

Before SM, forces held ‘the news’ about their doings but worked through intermediaries to reach audiences; now intermediaries are being supplanted. Diffusion reverses the traditional press officer’s approach, whose natural instinct was to control information, sometimes inhibiting it.

Radio picked up and talked about [police] tweet. So it’s morphing into other channels. They’ve got their information from it . . . [Y] ou don’t need the media . . . because they should all be following you . . . Tweet ‘Motorway is shut, avoid’ and people do it. (DI: 4)

One respondent declared, ‘we’ve become our own media agency, our own press. We are putting out our own stories . . . before we give it to the local press. [W]e are . . . very balanced and transparent’ (KB: 9). The last remark suggests awareness of the subversive aspect of these developments. Comms teams are negotiating hoary mass media issues, just as citizen journalism has eroded comfortable assumptions about source integrity.

We first dealt with fake news in 2011 riots. Somebody [was] saying that the local Primark is on fire, that was fake news, that was beginning to get momentum on social media, we were the credible, trusted voice that said ‘no, and here’s a CCTV shot of the front of the shop’. (EC: 7)

Stumbling into the role presents Comms teams with unanticipated challenges in enforcing standards and monitoring the effects of their increasing news function. It is moot whether the power of the press should accompany powers of arrest.

If Comms teams represent an alternative voice within forces, sourcing public views and sometimes direct involvement, one potential may be to gently remake a traditionally hierarchical organization.

Comms team don’t work 24 hours a day, but we’ve got a control centre that do, so we’ve trained . . . the switchboard. They can now look at social media in the middle of the night . . . We’re [also] doing a lot of work on a representative workforce reflecting our [diverse] population . . . So we’re [tweeting from] community halls [and] providing access to third parties or charities. (KT: 8)

This also leads to working across established roles. The public ‘don’t care whether it’s a Comms officer or a switchboard operator, they just want someone to go back to them with an answer’ (KT: 8). If an issue recurs in inboxes,

I will speak to the supervisor . . . There’s sometimes local community issues where people were contacting us complaining that they had spoken to an officer but [nothing happened] and they felt isolated. It reminded me of [the Fiona Pilkington case] . . . the effect of social exclusion, so we do try and get back to them. . . . it’s re-educating officers that haven’t been thinking like that. (KT: 8–9)

The need for timely response cut across established procedures. Customarily, public communications were carefully managed, with sign-off by senior officers for all but the entirely routine. Pressure to be first in breaking incidents works against this, making Comms less backstage.

It used to be you would need to make sure there was somebody on the scene or in the control room that was going to sign off all communication, that could take a really long time . . . their first thought is not a statement, their first thought is ‘what is going on, how do we keep the public safe?’ In social media you can’t wait for somebody to sign off your line. (DI: 6)

This is not least because searches and streams needed to be set up to check what others were saying. Communication is further diffused because Comms teams create local accounts managed by PCSOs who ‘provide the content, the whole aim is to show what’s going on in that area and if they’re walking about and can take a photo of what’s going on that’s far more effective’ (KT: 3–4). There is a hint that Comms can register as a separate driver at incidents, causing friction: ‘in an incident it is still thought of as a bit of a pain when people start responding [to Twitter posts about it]’ (DI: 6). Neighbourhood teams needed training and encouragement to adopt Twitter and Facebook. Older frontline officers wanted to get on with the job.

Limits to SM’s brief were recognized, including reporting discipline and complaint cases against officers.

It doesn’t work if we’re putting out [that] an officer has been disciplined or there’s going to be a [disciplinary] panel meeting . . . The public and journalists can attend. We don’t put it on social media because it attracts so many negative comments . . . We’re being transparent by putting it out to the media and on our website. (IG: 8)

Crime statistics and force inspections were also not suited, because ‘people can see through the stuff you’ve just done because you have to’ (DI: 5). Going from ‘needing improvement’ to ‘good’ at inspection was not ripe for SM but, once aired via other media, SM could ‘correct inaccuracies’ (DI: 3). The distinction between asymmetrical and symmetrical communication is helpful in framing the engagement agenda. There is a trade-off between engagement and being authoritative. Thus ‘we’ll always have a purpose for being involved in the conversation, to keep people safe, so it can never be symmetrical’ (EC: 8). However desirable, engagement cannot be elevated above policing’s core mission.

Discussion

The decentralizing affordances of SM reflect widespread de-bureaucratization initiatives by officialdom. Most Twitter communication involving Dutch police occurs via decentralized channels (Meijer and Torenvlied, 2016). This is a live issue with UK officers, who attempt to balance the medium’s informality and an authoritative tone by mostly using their rank and surname. Meijer and Torenvlied found a ‘hybrid organization of SM communications’. Bullock’s (2018) case study in an English force found that the claimed potential of SM to extend public participation in everyday policing was frustrated by organizational and cultural obstacles. In her analysis SM uptake and deployment are shaped by concatenated organizational, technological, individual, and cultural dynamics. This is a useful heuristic against which to consider the ongoing development of police SM communications, superseding the hopes and hype that marked the emergence stage.

SM may be mildly transformative within police organizations but evidence of effects on the police–public relationship is limited. A modest effect on legitimacy manifests itself amongst a small fraction of ‘interested citizens’. In Grimmelikhuijsen and Meijer’s (2015) representative sample of Dutch citizens a ‘negligible’ number engaged online with police. Twitter communications principally enhanced the transparency of police actions and policies. Whereas 35.8 percent watched TV documentaries about policing and 15.2 percent of the sample used Twitter, only 3.4 percent engaged with police tweets, and there was little evidence of SM providing information of direct operational value. However, reaching more citizens can enable direct participation, witness the examples noted here. Reach is also extended by 24/7 availability (Meijer, 2014), reflected in our respondents’ emphasizing responsiveness to out-of-hours contacts. Meijer (2014) suggests that forming new ‘virtual networks with citizens’ can seed ongoing participation beyond one-off incidents.

An index of reach is the extent to which SM messages are diffused, which is assessed by message forwarding. Tone/style as well as content affects the likelihood of forwarding police tweets. An informal tone/style increases the odds (Van de Velde et al., 2015), as does including URLs. Including Web addresses and hashtags increases diffusion (Suh et al., 2010), the effect being strongest for Web addresses. Our respondents emphasized both informal tone/style and providing links.

Another factor in message forwarding is user characteristics. ‘Interactivity’ (using mentions and replies) and ‘authorship’ (node centrality, where the sender is a regular information source, prompting audience retention) are key components. Account age and number of messages posted also predict diffusion – established, busy accounts indicate orientation to audience preferences (Marwick and Boyd, 2010). But message attractiveness is topic sensitive. Missing person appeals elicit sympathy and forwarding, traffic alerts less so. Message characteristics maximizing the probability of forwarding are, ordered by effect size, ‘send replies with URLs, include URLs, use mentions to show you are socially engaged, include hashtags to increase searchability, write longer tweets and send tweets in the afternoon/evening when more people listen’ (Van de Velde et al., 2015: 11).

Some themes presented here resonate with research on Web 2.0 affordances for promoting engagement with government institutions, but distinctive inflections arise – for instance, the obstacles that senior officers pose through naive aversion or enthusiasm, and that SM may to some extent alter the function and standing of Comms teams in the organization. Although this study suggests that SM’s increasing role in government communications now extends to the police, policing was a latecomer, and the present sample of Comms teams was alert to limitations and obstacles. Nevertheless, the interventions reported by respondents suggest the lines along which it is likely to develop, along with those that appear less promising. The findings also suggest that SM communication can displace more traditional media. Ultimately, the adequacy of police use of SM for communication purposes relies on the adequacy of police understanding of how social networks, interest groups, and community-based organizations function. That requires appraisal of vulnerabilities, public expectations, attitudes towards trustworthiness, and temporal variations in participation. These factors also come into play in respect of the standing of SM communication within the police organization itself.