PSA Conference paper

An examination of Facebook as a space for political talk through ethnographic study of citizen engagement during the 2015 and 2017 UK General Elections.

Sue Greenwood, Staffordshire University

In 2016, Mark Zuckerberg told his audience that Facebook is both a technology company and a media company – just not a traditional one. He told them:

Facebook is a new kind of platform. It’s not a traditional technology company. It’s not a traditional media company… We don’t write the news that people read on the platform. But at the same time, we know we also know that we do a lot more than just distribute news, and we’re an important part of the public discourse.[i]

Recently, we’ve seen just what “doing a lot more than just distributing news” might mean to Facebook, through the investigative work of Carole Cadwalladr and other journalists. Fake news farms; harvesting of data; political disintermediation. Facebook – whether directly or by omission – is doing a great deal more than just distributing news.

Facebook is not a traditional media company in that it doesn’t write the political narrative but, to yet unknown degrees, it may control it.

However, this paper focuses on that idea of Facebook as a new kind of platform. Something that hasn’t existed before, something very new and something very much a platform in the non-technological sense of that word – that of a space and opportunity to voice views.

In particular, it looks at that space delivered by Facebook within the context of the public sphere. By focusing on whether the social network provides a new platform for political debate.

The paper is based on research I conducted during the 2015 and 2017 UK General Elections for my PhD. While many studies have considered Facebook’s contribution to political debate by studying the role of the internet generally, or of social media broadly, my project focused on Facebook itself on the basis that its scale, reach and corporate ecology necessitates studying Facebook as a political actor in its own right.

Facebook is a platform that is actively used by over two billion people each month to chat, share content, organise events, flag locations, post pictures, thoughts and opinions, and to interact with people they largely already know. This is, again in Zuckerberg’s words, not just a communications medium but a “transformational” communications medium, able to change the way society is organised, to “give everyone a voice”.[ii] But a voice to do what, and to be heard by whom?

My research collected data during the 2015 UK General Election, supplemented by additional material gathered during 2017’s “snap” General Election, to look at how citizens engaged with each other in political talk on Facebook.

The mixed methods research included surveys alongside ethnographic study, and direct intervention via a case study. The methodology was influenced by a number of earlier studies. In particular, the work of Dan Jackson and colleagues in their 2013 study of political talk in online forums[iii], and danah boyd’s “deep watching” of teens’ talk on MySpace[iv].

By focusing on online spaces (political candidates’ public Facebook pages) and a time period (the month preceding a general election) when political debate might be expected to happen, the study assessed whether Facebook’s architecture encouraged or discouraged such debate, and whether this online social space is able to counter our offline social reluctance to discuss politics.

I sought to explore whether these online spaces – candidate’s Facebook pages – offered a 21st century version of Habermas’s 18th century coffee house and salons as a place where debate is expected, sought out, and encouraged.

The data gathered was in several forms. Stage one consisted of surveys targeting journalism, politics and humanities students who, because of their choice of study, might reasonably be expected to be more interested in political and civic issues.  Plus, a control group survey conducted with Games Design students, selected because choice of study did not suggest heightened interest in politics. All the participants answered the same questions around their engagement with Facebook and interest in news and politics.

Stage two data focused on the 2015 General Election, with Facebook feeds of candidates contesting four representative constituencies – Bristol West, Brighton Pavilion, Stoke-on-Trent Central, and Burton & Uttoxeter. These were monitored in the weeks leading up to the election.

In all, around 3,500 posts and comments were read, looking for evidence of political talk.

In addition, material from news feed on the day of election was sent to me by a number of participants, and I conducted interviews with some of the candidates.

Stage three was a bonus. The calling of a “snap” election by Theresa May in May 2017 provided an opportunity to look at changes in Facebook in the intervening two years, and to focus on another aspect – whether engagement in political discussion could be encouraged by posting particular material.

Basically, I was given access to the Facebook page of a new and largely unknown candidate and able to write tailored posts, watch how comments were managed by campaign volunteers, and track the effect of paying to boost posts to expand reach.

I should also say at this stage that there was another reason why I was able to re-engage with the research at the 2017 election. I’d missed my original PhD completion date of November 2016, because I’d taken a year out to write a book – this one.

It’s called Future Journalism: Where we are and where we’re going, published by Routledge, and it’s a very good book!

Aside from giving it a plug here, there is another reason for mentioning it which is that during my research for the book I interviewed a number of people at news organisations, including editors focused on expanding their publication’s reach on Facebook. So I also now knew more about posts “travel” on Facebook and how the platform’s algorithms might affect who sees – and therefore might be influenced by – content.

So, what did I discover from all of this data?

My 2015 surveys found a majority of young people locked in a “want to talk, daren’t talk” relationship with expressing political views to their Facebook friends.

Two-thirds of the young people surveyed said they were interested in and wanted to debate socio-political issues, andyet they routinely avoided commenting on issues on Facebook – even when moved by a campaign or news story.

However, much they might say they enjoy taking part in a good debate in the real world, and however much they might want to support friends who did talk about serious issues[v] on Facebook, they were not comfortable starting debates themselves on the social network. They were: “want to talk, daren’t talk”.

Daniel Miller and his colleagues[vi] found a similar reticence in discussing politics on Facebook in his ongoing anthropological survey “Why We Post”. In each part of the globe studied, they found that political talk was less likely to happen on social media than offline.

In 2015, I also noted a debate versus polemics split – with debate more likely on Green candidates’ Pages and least likely on UKIP candidates’ pages. Debate was also more likely among younger voters and women, and polemics – soap-boxing – more likely from older men, and on UKIP and Conservative candidate pages.

And I found that, across the thousands of Posts and comments I saw, and despite those posts being on political pages in the days before an election, and despite those posts largely coming from people who had shown interest in the election by choosing to visit a political page, there was actually very little political debate. Only 170 threads (5 percent) of the material read showed evidence of reasoned debate.

It should be understood that I was not simply looking for political talk, but for political debate. Debate that, as Schudson argued, was among people “acquainted by virtue of their citizenship… under norms of public reasonableness”, rather than simply social interaction.

Habermas saw the formation of public opinion as a cornerstone of the public sphere. Evolving from the “world of letters” of coffeeshop clubs, salons and the press, public opinion would put the state in touch with the needs of society[vii].

A Habermasian test of whether Facebook delivers a public sphere would therefore include the structure of the communication – that is “public” and “unfettered”; its quality (“reasoned”, “ideas”, “debates”); and its result (the “formation of political will”, the forming of “public opinion”).

This research was about identifying Facebook’s relationship to the UK public sphere in its Habermasian sense, and therefore sought to identify political talk that was public, participatory and reasoned debate rather than oratory; that circulated ideas or information with the aim of helping form opinions.

I therefore coded content in ways to distinguish between political talk that invited debate, and talk that was merely opinion sharing.

Posts and comments were coded as O – oratory or declarative view-sharing; D – debate, or S – simply expressing support for the candidate, or agreement to a comment made by someone else.

That five percent of reasoned debate, swamped as it was by oratory and simple support, felt like a surprisingly low figure, but what would be an equivalent comparison point?

While Verba and Nie found, in their 1987[viii] survey of US citizens’ political participation, that 28 percent of people said they had attempted to persuade others to vote for a particular party or candidate, that could as easily be via oratory as via debate.

And, while there has been research into the percentage of political talk in everyday conversation – this was not everyday, and by-and-large not conversation.

What I found particularly interesting in the 2015 ethnographic research, was how debate was conducted – the “public reasonableness” element, with signalling language to encourage participation.

As example, a thread on the Stoke-on-Trent UKIP Page began as oratory from Facebook user Jason about local councils, the EU, and ‘preferential’ treatment for migrants. But, rather than a racist rant, he ended his oratory with the conciliatory “Rant over” a signal of reasonableness – to which Rich responded, opening his comment with: “You’re right of course in some ways Jason but…”.

The two men were essentially on the same political side, but the exchange became a debate as Rich responded to Jason by trying to persuade him to become a stronger advocate for UKIP policies.

Jason would open a sentence with “I understand [however]”, Rich with “I agree [but]” and in doing so, both signal willingness to continue. Then, whether to challenge Jason, or having tired of the discussion, Rich veered from reasonable debate towards personal criticism, accusing Jason of “essentially spouting lefty lies” and returned to oratory. Jason ignored him and the discussion ended.

The almost apologetic sign-off to making a political point – such as “rant over” or “I’ll step off my pulpit now” was seen frequently. Women in particular would often sign off with emojis, X-kisses, or a jokey element. Signalling reasonableness and therefore an invitation to others to step in, but also saying: “it’s ok – politics bit over now and nothing scary happened”.

So, Michelle felt it necessary to temper her strong views (“if Labour get in omg God help us”) in consciously chatty language and a willingness to listen to opposing views.

Writing on her own Profile page she is happy to debate with her non-Conservative-voting “girlys” as she addresses them, because more people “should attend debates and understand politics”, but reassures them: “don’t want to get into a row over politics Hun you can vote who like.”

The etiquette was interesting – the women in this conversation called each other “Hun” and acknowledged the other’s opposing view – reassuring themselves that the friendship (at whatever level it was on Facebook) would not be affected by their debating from opposing political sides.

When another Friend joined the debate, and who also disagreed with Michelle’s political position, he eventually signalled an end to the discussion with an invite to his gig – an olive branch she happily accepted.

There is reasonable debate but fairly short-lived debate at that, perhaps because the participants are too quick to end discussion if it seems to be getting too serious, and therefore a risk to the equilibrium of the Friends group.

Facebook’s structure, as Miller noted in 2011[ix], encourages normativity with its restricted categories of content and direction of user actions. But also, as my student surveys had reinforced, there is a greater reluctance to debate politics online.

This normative behaviour is intended to maintain the equilibrium of the group with carefully emollient language – at least when a user is on their own Profile page and within their own friendship group – as Michelle was. However, having stepped out of their own group to visit a candidate’s page, there was more freedom to debate and declare. Albeit, mostly declare.

There were many similar examples of people apologizing ahead of making a political point, or tempering their views in order to mitigate risk to the group.

A post from one young voter starts with: “I’m not one to comment on my Political Views via social media but…”, before showing his hand holding a UKIP poster he’d torn down. Finally wrapping up with his own version of ‘politics bit over, nothing scary happened’ with the invitation:

“Now what’s everybody wearing to the Polling Station? I’m going with brogues, skinny black jeans, and a white grandad collar shirt.”

Whilst the 2017 data collection was focused around one candidate rather than watching what was going on in the four constituencies, what I did notice was tetchier, harsher language. Including candidates directly attacking each other and allowing, if not enabling, similar behaviour from supporters on their page.

A Conservative candidate who, in 2015, had told me he deliberately avoided contentious issues on Facebook because “people get turned off by ‘yahboo politics’”, in 2017 visited a rival’s Page to directly attack him.

However, what I also saw was people using Facebook to seek out spaces in which to see and join in political discussion with strangers. This willingness to engage in debate with strangers was particularly evident on Green Party MP Caroline Lucas’s page, and even more so after the election result.

Unlike the Pages of other winning candidates, Lucas’s Page showed hundreds of people – of all parties – coming together to debate the election result, rather than to simply congratulate her. Lucas’s page seemed to act as a magnet for hundreds of left-leaning voters, whether they voted Green or not, and whether they lived in Brighton or not, to come together in order to debate the election result and the political future.

The prompt to debate seems to have been a post by Lucas arguing for cross-party progressives to work together in opposition. But the debate on her Facebook page was dominated by people who were not in Lucas’s constituency. The question then becomes, why were they there?

By visiting or linking to Lucas’s page, it is reasonable to assume most of them were looking to engage with like-minded people. Being there would validate their own thoughts on the election; offer new insights, and the opportunity to debate in a space more likely to be reasonable. They were seeking debate with strangers, but with strangers who thought like them.

If one were looking for the roots of ‘Corbynmania’ in 2015, those passionate May 9th debates on Lucas’s Page would be a good starting point.

Aside from the changes in the political Left, what had also emerged during the two years between the elections was more awareness about how Facebook’s algorithms influenced what content, including political content, a user might see.

Returning to Facebook for the first time in 2017, I found the platform knew a great deal more about me that it had in 2015. Suggesting people I may know and delivering adverts that seemed based on the network being able to mine data from other areas, for example, my WhatsApp feed.

While that is touched on in the research, along with the investigations into the gaming of Facebook’s algorithms by partisan political groups, my 2017 work focused on whether debate is affected by the content posted and by how far the candidate engaged in debate on their page.

My 2015 data had shown that, of the 27 candidates across the four constituencies studied, all of the winning candidates and three out of four of the second-placed candidates were also  the ones who – personally or through their agents – were most active on Facebook.

My 2017 research therefore sought to look at whether the nature of that activity affected debate.

Further, how would paying to boost posts – the core of Facebook’s business model – affect the visibility and reach of that post and, in turn, engagement with it?

To wrap up I want to return to the starting point for the research – is Facebook part of the public sphere, or simply an additional public space in which political discussion may happen?

I’ve already spoken about the quality of the debate as being key – that reasoned and reasonable point – but Dahlgren, and others, see that interactional element as only one part, there also needs to be a structural and representational element.

In short, is the right sort of debate happening, and does Facebook’s ecology and architecture make that debate more or less likely to happen and, having happened, is that debate public and able to be seen by – or represented to – the state?

I found that Facebook met some of these parameters but not others. It is only sort-of public – its structure is a walled garden of software and architecture it controls, and yet makes available to an unknown number of third parties.

It links us with people we already know – our “trusted friends” – and yet encourages us to expand that network to a point where we no longer feel in control of who sees what we say, and therefore become less willing to debate.

It tells us its focus is on showing us material we seem to be interested in, from people or organisations we care most about, and yet its algorithms are easily subverted to show material we’re not interested in, or would not otherwise have come across.

Does that mean Facebook doesn’t meet the public sphere test? Or – and this is my argument – is Facebook so large, so influential, and so globally connected – and I’m not only talking abut its users here – that it has subverted it?

The Habermasian public sphere assumes a flattened perspective: people – media – state. Facebook is a massively global networked business experienced as a local-to-you space. Billions of bubble communities which influence each other, outside of state and media.

Facebook’s sphere of influence is who we know – the personal to the personal. This is a new public sphere of political influence, but one that redefines the word “public”.

It is bypassing traditional media’s filtering role between public and state. This is a sphere able to influence publics but not be public. Able to be manipulated by sinister third parties, and by teenagers making fake news videos, and controlled by algorithms prioritising paid content.

Whilst many of those things may be true of other media and social media, they are not Facebook. They are not an information source for one-in-six people in the world.

Facebook is not part of the Habermasian public sphere, it has disintermediated and replaced it.


[i] In a one-to-one video conference filmed with Facebook CEO Sheryl Sandberg and published to Zuckerberg’s Facebook page on December 21, 2016.

[ii] Zuckerberg’s letter to shareholders, reposted to (and accessed via) February 1, 2012.

[iii] Jackson, D., Scullion, R., Molesworth, M. (2013) ‘Ooh, politics. You’re brave’. Politics in everyday talk: An analysis of three ‘non-political’ online spaces’, in. The Media, Political Participation and Empowerment’, 205-220, Routledge, London.

[iv] danah boyd [correct] (2008), “Taken out of context: American teen sociality in networked publics”, University of California-Berkeley

[v] Overall, 65 percent said that if a friend made a good point about a serious issue on Facebook, they would usually Like it or reply.

[vi] Ongoing (at the time of writing) global anthropological study ‘Why we post’, led by Daniel Miller (, begun 2016.

[vii] Habermas, J (1989) ‘The structural transformation of the public sphere’, MIT Press, Cambridge, Mass.

[viii] Verba, S. and Nie, N.H. (1987) ‘ Participation in America: Political democracy and social equality’, University of Chicago Press

[ix] Miller, D. (2011) ‘Tales from Facebook’, Polity Press, Cambridge