A Longitudinal Analysis of Swedish Local Governments on Facebook: A visualisation of communication

The development of social media in the 2010s has transformed both the public sphere and our ways of interacting with each other and organisations. Authorities, especially local governments in close and daily contact with its citizens, are profoundly affected by such developments. As opposed to the written culture of governments, online social media are engrained in a visual culture. Today, local governments operate in a digital environment of e-services, e-government, and e-participation, and digital and social media are commonalities.

Previous research on interaction through social media in the Nordic countries has shown interesting results. For instance, Bonsón and colleagues (2015) measured the impact of content and media types across 75 local governments in 15 Western European countries. They found that Nordic and southern municipalities have the most engaged stakeholders. Swedish municipalities are diligent social media users, most have one or several Facebook pages, and some have used Facebook for more than a decade. These usage rates reflect the increase in social media use amongst Swedish citizens aged 9–79, which increased from 26 per cent in 2009 to 65 per cent in 2017 (Facht & Ohlsson, 2019). Today, 98 per cent of the Swedish population can access the Internet in their home (Facht & Ohlsson, 2019).

In a recent dissertation focusing on municipalities’ social media use in Sweden, Livia Norström (2019: 19) argues that social media in local government can have innovative use, by supplying service, external relations, and “access to knowledge beyond organisational borders”. Social media provides a hybrid medium through which local governments can inform, pursue dialogue, receive feedback, and monitor discussions (Rasmussen & Ihlen, 2017). Social media is also a part of a set of digital technologies that continues to be central in a discourse of development and efficiency within the public sector (Bertot et al., 2012; Norström, 2019). Local government communication and social media have been an intensive topic of research and discussion (see, e.g., Bertot et al., 2012; Bonsón et al., 2015, 2017; Guillamón et al., 2016; Lappas et al., 2018; Mergel, 2013; Miranda et al., 2018). Most researchers agree about social media's potential to reduce agency costs and support a network governance framework. However, it appears that most local governments use it as one-way communication and as an extension of existing channels, such as the government web page (Bellström et al., 2016; Norström, 2019; Magnusson et al., 2012). The Facebook communication of local governments is, therefore, typically described as resulting in high reuse of information from trusted institutional sources (Mergel, 2017). However, such information recycling should be seen as one of many tactics to maintain a presence in social media, a presence that requires continuous moderation and awareness of spin (Guillamón et al., 2016).

Research on local governments’ uses of social media consists of a patchwork of empirical studies with limited data capturing small moments in time. By looking back and reflecting on how local governments’ uses of social media have developed between 2009 and 2017, the purpose of this study is to paint a broader picture. More specifically, I pose three research questions:

RQ1 How has interaction developed over time in terms of likes, shares, and comments, and what seems to drive interaction for municipalities?

Like all organisations, local governments are “embedded in social environments and governed by structural preconditions”, and therefore, their social media practices are “enabled and constrained” by the social (Fredriksson et al., 2013: 184). Politics in Sweden, and especially at the local level, is characterised by strong centralised parties and high voter turnout, by international standards. Swedish municipalities have considerable autonomy and taxation rights and provide a vast array of services, such as schools, elderly care, local transport, housing, garbage disposal, and more.

According to Fredriksson and colleagues (2018), communication work in Swedish municipalities is conducted according to seven principles: organise and create a structure for communication work; position the municipality in relation to other municipalities; have routines for alerting in the event of a crisis; be uniform to appear consistent; interact with stakeholders to capture opinions; use web and digital services to serve citizens; and interact with journalists and the outside world to inform. These principles reveal the comprehensive nature of communication work in the public sector and point to its difficulties and tensions. For example, a fifth of the Swedish municipalities lack a journalistic presence, making information dissemination more challenging (Facht & Ohlsson, 2019). From the citizens’ perspective, this also means less insight and investigation of political and administrative decisions. Such challenges also exacerbate tensions that organisations must deal with, such as creating a uniform brand whilst maintaining transparency (Fredriksson et al., 2018; Fredriksson & Pallas, 2013). In short, communication work interacts with laws and regulations (for instance, the principle of openness and secrecy legislation) that make it more difficult for municipalities to communicate. With the rise of social media, municipalities must adapt to these preconditions in unison with a system where one's personal information is currency, and the control over – and predictability of – messages might be nullified by commenting and sharing on a large scale.

Local government communication departments usually do not have their own goals. Instead, they plan their activities based on political priorities and support other departments in activities to fulfil (politically set) priorities and goals. However, not all local governments have the same financial and organisational means to pursue their goals. Local governments vary widely in both the tax base and human capital they have available, which influences their political priorities and thereby shapes the communicational priorities. This variety also applies to the opportunities to use social media. Lidén and Larsson (2016; see also Larsson, 2013), for instance, note that municipal size and use of social media are positively correlated. In the Swedish context, Bergquist and colleagues also point to differences in how authorities use social media in different ways; for example, the police using it for “flexible interaction” by utilising social media's various capabilities, or the Swedish Social Insurance Agency creating a more “stable and well-defined” customer service channel (Berguist et al., 2017, 870; see also DePaula et al., 2018). Since Swedish local governments perform a comprehensive set of services, this will presumably lead to social media being used both as a means for flexible interaction and as a customer service channel.

Several frameworks have been developed to capture municipal use of Facebook, as well as the various levels of engagement (e.g., DePaula et. a., 2018; Guillamón et al., 2016; Bonsón et al., 2015; Mergel, 2013). One way to characterise social media use in the public sector is through a framework of push, pull, and networking (DePaula et al., 2018; Mergel, 2013). Though previous research concludes that the level of engagement depends not only on strategies and media usage, but also on networking with the community (Picazo-Vela et al., 2016), due to the nature of this particular study, my focus is on push and pull. Networking, which captures collaboration with citizens in issue networks or communities and co-design of services (Mergel 2013), requires more qualitative approaches, and preferably in-depth local knowledge on issue networks and governance schemes.

Pushing aims to increase transparency by informing and educating citizens through one-way communication. Studies show that the push category is most common, particularly for information provision (Gao & Lee, 2017; Reddick et al., 2017; DePaula & Dincelli, 2016; Mossberger et al., 2013) and marketing activities (Bellström et al., 2016; Bonsón et al., 2015; Magnusson et al., 2012). Mergel (2017: 490) argues that governments are prone to push tactics due to the “risk-averse” context in which they operate, and therefore, they often recycle information already published on their website or by trusted institutional sources. This strategy helps avoid posting incomplete information, which can erode citizens’ trust in government, even in less severe matters (Wukich & Mergel, 2016). Furthermore, though expressing a reservation for cross-consumption of information on multiple platforms, Porumbescu (2016) suggests that users are more receptive to less-detailed information, favouring social media as a medium for pushing information. Another aspect contributing to the popularity of social media for pushing information, according to Fredriksson and colleagues (2018: 21), is that municipalities perceive social media as “fast, easy and cheap”. While pushing information is the most common tactic DePaula and colleagues (2018) found that 45 per cent of all posts in their study were symbolic, and a large share of these posts (21%) were acts such as “thank you” or “congratulations”. Thus, they raise the importance of understanding the “social” in social media through a ritual model of communication (see Carey, 2008), not as a mere transmission of information from a sender to a receiver.

The push approach mainly follows what seems like a tradition of e-government and information provision (Hanssen, 2007; Torpe & Nielsen, 2004). There is a relatively linear relationship between governmental adoption of social media and delivery of online democratic services via official homepages, making e-government a trend that may be hard to opt out of (Lidén & Larsson, 2016). The relationship is also apparent in how local governments share links on their Facebook pages and through the guidelines that dictate how they should relate to each other.

In contrast to the one-way communication of push, pull is used to stimulate the interaction and engagement of citizens in deliberation, consultation, and feedback (Mergel, 2013), for instance, through encouraging language (Gao & Lee, 2017). E-participation of citizens seems to result in governments posting more information via social media, as the acts of participation often require government responses (Guillamón et al., 2016). Given the broad activities and services Swedish local governments provide and the party-centred nature of local politics in Sweden, pull messages often contain a comprehensive set of events and issues. In terms of interaction, large amounts of likes, shares, and (positive) comments may serve as an indication of how an issue or policy is received. Similarly, the absence of any reactions can be interpreted as a sign of disinterest.

Adapting posts to include more visual material may be vital to pull strategies as a way to increase user interaction (Lappas et al., 2018; Lev-On & Steinfeld, 2015), and may thus be a strategic tactic for public relations work (Fredriksson et al., 2018). For instance, Rasmussen and Ihlen (2017: 12) note in a literature review on social media, risk, and crisis that social media can be used proactively as a “listening post” and a space where organisations can float suggestions and engage with stakeholders and risk bearers. Lastly, the ability to pull, listen to citizens, and thus increase interaction should be understood as deeply connected to “rhetorically configur[ing]” social media usage “to establish stronger bonds of friendship with the next generation of media-savvy citizens” (Isaksson & Jorgensen, 2018: 126).

Studying the Internet and social media is, as Karpf (2012: 652) puts it, “unsettling”. It requires hitting a moving target – analysing phenomena as they change. The data collected from 38 Swedish municipalities’ Facebook pages was initially meant for use in a pilot study conducted before material from all 290 municipalities was collected. As a consequence of the Cambridge Analytica incidents, however, Facebook made communication with their application programming interface harder, thus limiting further access to municipal Facebook data.

This study's data was gathered by communicating with Facebook's application programming interface using a python-script specifically configured to mine the posted content and interaction figures of a given page. The script only retrieved descriptive statistics and not the comments, images, or videos posted, which has both advantages and disadvantages. For instance, the source of the collected data has been accessible except for Luleå municipality, who reshaped their Facebook page shortly after the data collection, making it impossible to follow the mined links back to the original post.

It is often preferable to physically store the digital material should it be removed or disappear, as often is the case. At the same time, such a collection involves more far-reaching ethical difficulties than collecting descriptive statistics, as samples often would include posts and comments from citizens.

The sample was randomly selected from a list of all municipalities and controlled for geographical and demographical markers, such as municipal size and geographic location, as well as municipality classifications set by the Swedish Association of Local Authorities and Regions (2016) (see Appendix 1 for full details of the municipalities’ data). The sample thus contains data from both early and late adopters, and during the studied period, they most likely have developed and changed their experiences and knowledge. There have also been political shifts in power and new developments that inevitably have affected the content and posting habits. The sample also includes three municipal place-branding Facebook pages: Jönköping, Luleå, and Halmstad. These pages differ from the “official main pages” not in their ownership, but rather in their focus. Place-branding pages focus more exclusively on aspects of marketing municipalities as places to live, work, and visit.

Data collection ran from the day the respective municipality made their first post until data collection ended on 15 September 2017, and resulted in 85,663 posts. Of these posts, 181 had a date preceding the Facebook page's existence and were excluded from the analysis.

From 2011 onwards, Facebook allows its users to add events on their timeline to chronologically order life (see Van Djick, 2013). For instance, Ystad municipality added a post from October 1910 on their timeline to highlight their first cinema's opening.

During the time this dataset was posted, new platforms such as Instagram and Snapchat gained momentum, affecting how people think of and use social media. Furthermore, new features, such as autoplay and focus on video, have during this time changed the functions of Facebook's software. Another limitation is that many posts contained images, links, and text in the same post; the classification here reflects only how individuals actually posted. The data also includes the total number of reactions to the ascribed post and the number of comments, shares, and likes. To retain continuity in the interaction measurements, I have excluded the reactions introduced by Facebook in 2017. Further, the dataset does not distinguish media type (for instance, video) for a post type such as events – they are all treated as media; however, it is crucial to know that an event can appear as a video or a photo. The raw data from each municipality was exported to Excel and then merged, providing the bulk of the descriptive statistics for RQ1 and RQ2.

For RQ3, I exported all links and used a text editor, TextWrangler, to isolate the external links from every municipality. I used a semi-automated approach by instructing the editor to select similar strings – such as “goteborg.se” – to categorise the links. 38,970 links were processed, but one fourth – many of them so-called short links like “bit.ly” – required a workaround and manual coding. To the extent possible, I traced broken links through Google, yet some links were untraceable, making the total number of links 38,076. To keep the categories of links at a manageable size and still be able to discern issues of information recycling (Mergel, 2013; Wukich & Mergel, 2016), I categorised links on sectorial considerations and organisational forms, such as NGO, commercial companies, governmental agencies or companies, and interest organisations. I also divided links into categories of local, national, and international news media, public service and magazines, and other social media to frame linking in a broader network of actors. All sorted links were exported to Gephi – an open-source visualisation and exploration software – for network analysis and visualisation and will be presented in the next section (see Figure 2).

This study aimed to explore nine years of Facebook data from Swedish local governments. I asked how interaction and media use developed over time, what kind of links local governments shared, and whether they indicated high information recycling levels. What is most striking during this period is how Facebook use becomes an ordinary life practice for many citizens (Davidsson & Thoresson, 2017) and how it reflected a substantial increase in municipal Facebook activity. The municipalities’ posting habits indicated that they were embracing a more visual approach, with images and videos gaining primacy over texts and links, a development consistent with findings in previous research (e.g., Bonsón et al., 2015; Lappas et al., 2018). Image and video content is an essential driver for interaction, which social media managers surely have noticed; however, cross-posting from other social media platforms like YouTube and Instagram was rare and often only accessible via a link through the municipalities’ websites.

The many links to municipal web pages suggests substantial recycling of information, indicating high levels of pushing information (Mergel, 2017; Wukich & Mergel, 2016). Furthermore, a close reading of policy material indicated a common view that information on Facebook should always have its equivalent on the homepage. A central finding is that Facebook had a remediating function and, at times – especially from 2010 to 2013 – Facebook pages could even be seen as extensions of the homepages (see Lidén & Larsson, 2016). This result is not surprising, as many municipalities are service-oriented organisations and have a responsibility to reach all their residents with important societal information (Fredriksson et al., 2018). Social media provides ample opportunities to reach “the unreachable”, anchoring its role as a complementary channel to informational hubs such as a government web page. We must also understand high levels of push information in the context of the absence and decreased resources of local news media, resulting in a more extensive – and problematic – responsibility being put on local government organisations to report “the news”.

In line with previous research (e.g., DePaula et al., 2018; Gao & Lee, 2017; Reddick et al., 2017; Mossberger et al., 2013), many posts seemed to be push messages about events and meetings or service issues like water leaks. Facebook, as a medium, perhaps shows its most significant strength in these latter – and what may be perceived as more trivial – matters; updating citizens via Facebook is easy, with instant access for citizens in a space were they already direct much of their attention (Davidsson & Thoresson, 2017). Push messages also included ritualistic communication geared towards identity, branding, and communal discourses and symbolical acts (DePaula 2018). As exemplified by the earlier discussion of the red moon post, push communication in a ritualistic configuration appealed to a sense of shared community and local pride, thus drawing citizens together (Carey, 2008) and seeking their approval (Van Dijck, 2013). These messages were often filled with visual representations of local symbols such as historical sights, city centres, or other beloved local monuments or places. Following Van Dijck (2013), we can also understand these posts as part of the construction and popularisation of a self-image. Facebook's software may help move this image to an external public by the inherent logic of lifting popular items (Bucher, 2012; Olsson & Eriksson, 2016). For many municipalities, Facebook and other social media are important channels for positioning themselves against the outside world and towards other municipalities. For many, and especially the demographically smaller municipalities, positioning is increasingly an existential issue – where to endure and thrive depends on the ability to attract visitors and human capital, and by extension, a tax base to finance welfare.

To conclude, social media, and particularly Facebook, seems to continuously play an important role for municipalities in more and more areas of their communication. Today, the chat function provides a fast and easy customer-service channel, and targeted ads give new opportunities to pinpoint hard-to-reach audiences and spread the image of the organisation across time and space. In recent developments, the Covid-19 crisis in 2020 has highlighted the value of using social media as a channel for informing and alarming citizens in a space where they already direct much of their attention. Further research should take more in-depth, longitudinal, and qualitative approaches to how local governments use social media as a space for discussions and listening, and to the role it may play in the local political system. There are ample opportunities for both politicians and public servants to propose suggestions and listen to the discourse. In the context of a larger political system, social media – and in particular centralised Facebook pages with large followings – can serve an important epistemic function if they are taken seriously as a forum for discussion and listening. In the end, it is a question of finding traces where people leave them.