Evolution, not revolution
Facebook mostly recreates the traditional campaign, only louder.
Research by Dr Tristan Hotham on how political parties, leaders and campaigns use Facebook and other platforms, and how audiences respond: engagement, comment sections, targeted advertising, populism and elections in Britain, Europe and the United States. Every paper is open access, with a DOI and aggregate data where the terms of collection allow.
Research map
Each paper placed where its data comes from. Choose a place to show its papers in the list below. Papers on the UK as a whole circle Britain; Scotland, Wales and London have their own markers.
Number of papers on each. A paper that compares countries counts in each one.
Quick answers: Do populist parties do better on Facebook? · What is a satellite campaign? · On Facebook, do party leaders or party pages matter more? · Can Facebook engagement predict elections? · How did British parties use targeted Facebook ads at the 2019 general election? · Did Leave win the EU referendum on Facebook by being more negative? · Does Facebook reward the political right? · What is Janus-faced campaigning?
No papers match that search.
Every party whose Facebook pages appear in the research, plus the campaigns studied in their own right. Colours show party family and a white dot marks parties the PopuList classes as populist. Tap a tile to see the papers about it, or search by name.
Party families follow ParlGov and populism follows the PopuList. Leader pages are listed under their party. The list holds party-level information only: no posts are shared.
Who actually pays for British party advertising on Facebook and Instagram, and through which pages? I use Meta's public Ad Library Report for the United Kingdom, which lists every advertiser running ads about social issues, elections or politics from October 2019 to October 2026: 90,565 pages that spent about £252 million on 1.51 million ads.
Finding: Party pages took only 8.3% of political ad spending on Meta in Britain; Labour put 38% of its spend through devolved, regional and local pages, the Conservatives 24%.
AbstractPDFSummaryParties and their leaders run Facebook pages that attract audiences without payment, and they also pay Meta to put ads in front of people. Do they pay for the pages that people already engage with?
Finding: Among party pages, ad spend tracked organic engagement (rank correlation 0.76); leaders' pages earned most of the engagement but got little of the money.
AbstractPDFSummaryIs a British party's Facebook audience built around its leader or its party page, and does either track the polls? I use 89,110 posts from the pages of eight parties and eighteen of their leaders between June 2010 and October 2020, which drew 218 million interactions, and match them to monthly averages of 2,767 national voting-intention polls. Leaders carried very different shares of their party's engagement.
AbstractPDFSummaryBritish parties campaign on Facebook through two kinds of page: the party's own page and the page of its leader. Do the two do the same job? I compare 1,208 hand-coded posts from the Labour and Conservative party pages and the pages of Ed Miliband, David Cameron, Jeremy Corbyn and Theresa May in the month before the 2015 and 2017 general elections, giving four pairs of pages.The division of labour was consistent.
AbstractPDFSummaryWhat did British parties say in their targeted Facebook ads, how often did they say it, and what did they spend?
AbstractPDFSummaryParties do not only tell voters what they will do. They also show voters who they are for. Kreiss, Lawrence and McGregor call this political identity ownership: the work a party does to be seen as the natural home of particular groups of people.
AbstractPDFSummarySatellite campaigns, organisations that campaign for a party without being part of it, became a visible part of British elections in 2017. The best known was Momentum, formed to support Jeremy Corbyn's leadership of the Labour Party. I ask what Momentum's Facebook page did that Labour's own pages did not.
AbstractPDFSummarySatellite campaigns, organisations that campaign for a party without being part of it, are now a familiar part of British elections. We know less about what they do between elections.
AbstractPDFSummaryThe 2014 Scottish independence referendum transformed the Scottish National Party. Its membership roughly quadrupled, and eight months later it won 56 of Scotland's 59 Westminster seats. I ask what happened to the SNP on Facebook after the referendum, comparing it with the UK-wide parties and with Plaid Cymru, the other main nationalist party in Great Britain.
AbstractPDFSummaryWhen a crisis strikes, does political attention turn from parties to government? I test this on Facebook in Britain using 274,047 posts from 437 political pages tracked by CrowdTangle between June 2018 and October 2020: government and Parliament pages, party pages and politicians' pages. Before the pandemic, government pages drew 5% of the interactions on government, party and politicians' pages combined.
AbstractPDFSummaryBetween the announcement of the EU referendum on 20 February 2016 and polling day on 23 June 2016, the Leave and Remain sides fought a large part of their campaign on Facebook. I analyse 17,657 posts from 92 UK pages, covering the official and unofficial campaigns and the main party, leader and MP pages, which together drew 33.5 million interactions.
AbstractPDFSummaryEngagement on politicians' Facebook pages is often read as a sign of public interest. I ask how closely it tracks two free measures of public attention, Wikipedia pageviews and Google searches, and whether it moves before or after them.
Finding: Facebook engagement moves with public interest rather than ahead of it, and is not a measure of support.
AbstractPDFSummaryWhich kinds of campaign post draw the most likes, comments and shares? I link a hand-coded dataset of 1,205 Facebook posts from the Labour and Conservative party pages and the pages of David Cameron, Ed Miliband, Theresa May and Jeremy Corbyn, published in the month before the 2015 and 2017 UK general elections, to the interactions each post received.
Finding: Many content codes were linked to engagement, but most were not reliably reproduced by a second coder.
AbstractPDFSummaryBritain First built a large following on Facebook before the platform removed its pages on 14 March 2018. The page and its post-level record are no longer publicly available. I analyse an archive of 3,505 posts published between 25 October 2013 and 12 March 2018, extracted two days before the ban, which together attracted 32.8 million reactions, comments and shares.
AbstractPDFSummaryJeremy Corbyn's election as Labour leader in September 2015 brought a surge of new members and a reputation for an unusually energetic, and unusually combative, online following. I test what changed in the place where that following met the party most directly: the comment section of Labour's own Facebook page.
AbstractPDFSummaryThe 2016 London mayoral election was contested by Sadiq Khan, who became the city's first Muslim mayor, and Zac Goldsmith. Religion and extremism became themes of the campaign. I compare the comment sections of the two candidates' Facebook pages from January to mid-April 2016: 36,087 comments on 701 posts, after removing duplicates.
AbstractPDFSummaryDo populist parties get more attention on Facebook than other parties, and is that true everywhere? I use 669,760 posts from 112 official party pages in 17 European countries, posted between January 2010 and September 2020, with 186,939 posts from 74 party leaders as a second sample.
AbstractPDFSummaryRadical-right parties often draw more engagement on Facebook than their share of the vote would suggest. Does that advantage tell us anything about where their support is heading? I compare three established radical-right parties in Scandinavia: the Sweden Democrats (SD), the Progress Party (FrP) in Norway and the Danish People's Party (DF).
AbstractPDFSummaryPodemos and Vox were both presented to the press in Madrid in the same week of January 2014. Podemos won 8% of the vote at the European election four months later and 21% with its regional allies at the general election of December 2015. Vox won 0.2% in 2015 and 2016, then 11% at the Andalusian regional election of December 2018 and 15% at the general election of November 2019.
AbstractPDFSummarySinn Féin's first place in the Irish general election of February 2020 surprised most observers. The party had lost half its council seats in May 2019 and polled at about 14% in October and November. I ask whether Facebook saw the surge coming.
AbstractPDFSummaryEurosceptic parties across Europe might be expected to have drawn attention from the United Kingdom's vote to leave the European Union on 23 June 2016. I test whether engagement with the Facebook pages of Eurosceptic parties and leaders in 19 other European countries changed around the referendum and later Brexit milestones, using CrowdTangle data.
Finding: Posts about Brexit or the UK rose to about 9% across Europe after the vote, but Eurosceptic pages drew no extra engagement relative to their rivals at home.
AbstractPDFSummaryCrowdTangle closed in August 2024, but many archived exports of Facebook posts remain in use, and each row carries an "Overperforming Score" that compares a post's interactions with an expected value. I examine what this score measures in 856,699 posts from 186 party and leader pages in 20 European countries, 2010 to 2020, exported in October 2020.
Finding: In archived exports the Overperforming Score ranks posts within a page almost exactly as raw interactions do; researchers should count interactions instead.
AbstractPDFSummaryWhen a faction or a prominent figure leaves a party to found a new one, does the party's online audience follow the breakaway or stay with the parent? Screening splits in 20 European countries in CrowdTangle data on 1,712 Facebook pages, I found ten, in seven countries from 2011 to 2019, in which the parent's page and the breakaway's pages (new party or departing figure) are observed around the split.
Finding: In ten splits, nine breakaways gained share, but the typical breakaway held only about 5% of national interactions afterwards.
AbstractPDFSummaryDo short posts draw more engagement on social media than long ones? I examine this using 856,220 Facebook posts by 112 party and 74 party leader pages in 20 European countries, 2010 to 2020, collected through CrowdTangle.
Finding: Posts with no text drew about 42% fewer interactions; once a post had text, its length made only small differences.
AbstractPDFSummaryDid the 2015 European refugee crisis show up on radical right parties' Facebook pages before it showed up in opinion polls? I compare 15 populist radical right party and leader pages in nine European countries with 68 other party and leader pages in the same countries, month by month from July 2014 to June 2016, using posts collected through CrowdTangle.
Finding: Radical right pages drew about 47% more interactions per post than rivals at home in September and October 2015, but Facebook did not clearly run ahead of the polls.
AbstractPDFSummaryA common claim is that Facebook steadily reduced the organic reach of pages. CrowdTangle does not record reach, but it does record the number of page likes at the time of each post, which allows engagement to be measured per follower.
Finding: Engagement per follower fell by 25 to 35% from 2016 to 2018, then recovered to about its 2016 level by 2020; there was no sudden drop at the News Feed changes.
AbstractPDFSummaryA party that posts a link on Facebook chooses where to send its followers. I classify the destinations of 410,308 link posts made by 112 party pages and 74 party leaders' pages in 20 European countries from 2010 to 2020 by the domain of each link.
Finding: Links to parties' own sites fell from 58% of link posts in 2010 to 39% in 2020 while news links rose from 17% to 36%, and news links drew the most engagement.
AbstractPDFSummaryWhen engagement with a party's Facebook page rises or falls from one month to the next, is that part of a Europe-wide movement, a national one, a movement shared by parties of the same family across borders, or specific to the page? Using 14,510 page-months from 173 party and leader pages in 20 European countries, January 2013 to September 2020, collected through CrowdTangle, I split the variance of each page's deviations from its own trend into these four parts with nested fixed effects.
Finding: About 26% of monthly swings in engagement were national and 11% Europe-wide; party families, including the radical right, did not move together across borders (3%).
AbstractPDFSummaryOnline attention is often described as a world of hits, in which a few items take most of the attention. Is the engagement that political parties receive on Facebook driven by a few viral posts, or spread across the many posts they make?
Finding: Engagement was concentrated but not extreme: the top 1% of posts drew 8% of interactions in the median page-year.
AbstractSummaryPDFCan the engagement parties receive on social media tell us how they will do at the polls? Studies of Twitter have reached mixed and often optimistic conclusions, but parties' own Facebook pages, where supporters respond to what parties post, have rarely been tested across many elections.
Finding: Engagement shares tracked vote shares but added little once the previous result was known.
AbstractSummaryPDFThe personalisation of politics suggests that leaders increasingly overshadow their parties, and social media might be expected to speed this up. I compare the Facebook engagement received by party leaders' pages with that received by their parties' pages, for 53 leader and party pairs from 44 parties in 11 European countries between 2010 and 2020, using posts collected through CrowdTangle and counting a leader's page only while that person led the party.
Finding: Leaders' pages drew 43% of combined engagement on about a third of the posts, with no sign of rising personalisation.
AbstractSummaryPDFOnline attention to parties is often read as a sign of their support. I compare two kinds of attention to 112 European parties in 17 countries from July 2015 to September 2020: views of their Wikipedia articles, in the local language and in English, from the Wikimedia pageviews data, and interactions with their Facebook pages, from CrowdTangle.
Finding: Wikipedia views tracked vote shares more closely than Facebook engagement, but neither beat knowing the last result.
AbstractSummaryPDFIn the mid-2010s Facebook promoted native and live video, and news organisations famously "pivoted to video". Did political parties follow, and did it pay?
Finding: Parties took up video while it drew extra engagement, and kept posting it after the premium had gone.
AbstractSummaryPDFA party's Facebook audience is the base on which its reach and engagement rest, yet little is known about how such audiences grow. I track the number of page likes of 112 party pages and 74 party leaders' pages in 20 European countries month by month from 2016 to 2020, using the counts recorded with each post collected through CrowdTangle.
Finding: Followings grew several times faster around elections, though most growth still came slowly in ordinary months.
AbstractSummaryPDFEuropean Parliament elections are often described as second-order national elections, in which less is at stake and parties and voters invest less. Does this hold for parties' campaigns on social media and the response of their audiences?
Finding: European Parliament campaigns were real but second-order: a smaller surge in posting and a much smaller response.
AbstractSummaryPDFWhen do political parties post on social media, and does the timing of a post matter for the engagement it receives? I convert the timestamps of 856,696 Facebook posts by 112 party pages and 74 party leaders' pages in 20 European countries, collected through CrowdTangle between 2010 and 2020, to local time.
Finding: Parties posted in working hours, but evening posts drew about 10% more interactions than midday ones.
AbstractSummaryPDFParties often change leader in the hope of a fresh start, and in an age of personalised politics a new leader might be expected to bring new attention. Is a change of leader followed by a change in the engagement a party receives on social media?
Finding: A new leader brought a moment of attention, but on average no lasting new audience.
AbstractSummaryPDFElection campaigns are the busiest time for parties on social media, but little is known about what happens to their audiences once the votes are counted. I follow 95 party pages on Facebook through 229 party-elections in 16 European countries between 2010 and early 2020, using posts collected through CrowdTangle, comparing each page in the weeks after a national election with the weeks before the campaign.
Finding: Engagement peaked in the campaign, fell about 40% afterwards, and anger rose on the pages of parties left in opposition.
AbstractSummaryPDFDoes social media level the field between large and small parties, or do the parties that win votes also win attention online? Research on this question, framed as equalisation against normalisation, has mostly studied what parties do online rather than how audiences respond.
Finding: Engagement followed votes less than proportionally, and the largest party rarely had the most engagement.
AbstractSummaryPDFNew parties are often said to do well on social media, which offers them an audience without the organisation and media access that established parties have. Do new parties' Facebook posts draw more engagement than those of established parties?
Finding: The evidence for a general newness premium was weak: new parties did not reliably out-perform established ones.
AbstractSummaryPDFParties decide how often to post on social media, and a common worry is that posting more spreads their audience's attention more thinly. I use 856,699 posts from 112 party pages and 74 party leaders' pages in 20 European countries, posted between January 2010 and September 2020 and exported from CrowdTangle, to ask how the engagement a page receives changes when it posts more.
Finding: Engagement rose almost in proportion to posting across months, but posts bunched within a day drew less.
AbstractSummaryPDFOn 11 January 2018 Facebook announced that its News Feed would favour "meaningful social interactions" between people and show less public content from pages, which it warned might see their reach fall. Internal documents later suggested that European parties felt the change pushed them towards more negative content.
Finding: The 2018 News Feed change did not cut engagement, but for about a year audiences shared party posts more.
AbstractPDFSummaryWomen in politics receive more hostility online than men, at least in some settings and on some measures. Most of the evidence comes from the text of messages on Twitter.
Finding: Women leaders drew half the angry reactions of men pooled, but no detectable gap once party and position were held constant.
AbstractPDFSummaryFidesz won every Hungarian national election from 2010 to 2022, and from 2015 migration was central to its campaigns. I use 66,587 posts from the Facebook pages of seven Hungarian parties and two party leaders, Viktor Orbán and Ferenc Gyurcsány, from January 2014 to September 2020, collected through CrowdTangle, to ask three descriptive questions.
Finding: From 2015 to 2019, 31% to 41% of Fidesz party page posts mentioned migration each year, against under 4% on Viktor Orbán's own page.
AbstractPDFSummaryThe Social Democratic Party (PSD) won 45% of the vote in Romania's parliamentary election of December 2016, more than the National Liberal Party (PNL) and the new Save Romania Union (USR) together. I use 36,922 posts from eight Romanian political pages, 2011 to September 2020, collected through CrowdTangle, to compare engagement on the three party pages with the vote.
Finding: In the month before the 2016 election the PSD page drew 7% of the three party pages' interactions against 61% of the three parties' vote.
AbstractPDFSummaryFinland's three largest established parties won about two-thirds of the vote cast for the parties studied here, but drew well under half of their Facebook engagement. I use 38,729 posts from eight Finnish party pages, January 2010 to September 2020, collected through CrowdTangle.
Finding: From June 2015 Finland's three largest established parties drew 44% of interactions against 67% of the vote, and engagement did not anticipate who would gain votes.
AbstractPDFSummaryPortugal's two main parties, the Socialists (PS) and the Social Democrats (PSD), won 70% of the vote among the parties studied here in 2019, but drew much less of the engagement with Portuguese party and politician pages on Facebook. I use 153,649 posts from 12 party pages and 63 politicians' pages, collected through CrowdTangle, and compare each party's share of interactions in the 30 days before the 2015 and 2019 national elections and the 2019 European election with its share of the vote.
Finding: Before the 2019 election the PS and PSD drew 43% of interactions against 70% of the vote, while small parties drew 2.5 to 10 times their vote share.
AbstractPDFSummarySwitzerland is governed by a collegial Federal Council drawn from the main parties, and some of its members run their own Facebook pages. I use 11,126 posts from eleven Swiss pages, those of the SP, FDP and SVP, five party presidents and three federal councillors, from May 2015 to September 2020, collected through CrowdTangle.
Finding: In March to May 2020, Alain Berset's page drew 50% of all interactions on eleven Swiss pages, more than the three party pages combined.
AbstractPDFSummaryGoverning parties tend to lose votes, and junior coalition partners often lose more. Does governing also cost parties attention on social media?
Finding: In government, a party's posts drew about 18% fewer interactions than the same party's posts in opposition.
AbstractPDFSummaryCan Facebook engagement tell us who will win an election? I compare parties' shares of interactions on their Facebook pages in the 30 days before 31 national elections in 15 European countries between 2011 and 2020 with their vote, their final polls and their previous result, for 172 party-elections.
Finding: As a forecast, engagement was poor: 9.6 points of error against 1.8 for the final polls.
AbstractPDFSummaryDoes the state of a country's democracy show up in how people engage with its parties on Facebook? I use 184,317 posts from 35 official party pages in Finland, Lithuania, Portugal, Romania, Slovenia and Switzerland, posted between 2010 and October 2020, with 60.6 million interactions, collected through CrowdTangle.
Finding: Across six countries, no V-Dem democracy indicator was clearly related to party engagement.
AbstractPDFSummarySurveys found that trust in governments rose across Europe when the first Covid-19 lockdowns began. I ask whether attention on Facebook moved in the same direction: towards the pages of governing parties and heads of government and away from the opposition.
Finding: The rally on European political Facebook was real but modest, short and hard to separate from noise.
AbstractPDFSummaryHow much of a party's Facebook engagement comes through its leader? I use 33,958 posts from the pages of Sweden's eight parliamentary parties and seven of their leaders, from January 2015 to September 2020, collected through CrowdTangle.
Finding: Åkesson's page drew 41% of Sweden Democrat engagement in 2015, but only 13% to 16% from 2018 to 2020.
AbstractPDFSummarySpain held four general elections between December 2015 and November 2019, as the two-party system gave way to five national parties. I use 74,884 posts from the Facebook pages of seven Spanish parties and seven of their leading politicians, from January 2015 to September 2020, collected through CrowdTangle, to ask two descriptive questions.
Finding: On Facebook, the parties that led Spain's governments were the quietest.
AbstractPDFSummaryBelgium's federal election of 26 May 2019 saw large gains for Vlaams Belang on the Flemish radical right and for the PVDA-PTB on the radical left. I use 95,373 posts from the Facebook pages of 121 Belgian politicians, mostly members of the 2019 Chamber of Representatives and federal ministers, from January 2014 to September 2020, collected through CrowdTangle.
Finding: Vlaams Belang politicians' pages drew 48% of interactions in the month before the 2019 election, against 12% of the vote.
AbstractPDFSummaryIn July 2019 New Democracy won a parliamentary majority in Greece with 40% of the vote, ending four and a half years of SYRIZA government. I use 31,532 posts from the Facebook pages of five Greek parties and four of their leaders, from January 2015 to September 2020, collected through CrowdTangle, to compare engagement with the vote and to follow the two main leaders' pages through the change of government.
Finding: In 2019 the bigger Facebook page belonged to the party that lost.
AbstractPDFSummaryBetween 2016 and 2019 the French party system was redrawn. En Marche won the presidency and a majority within two years of its founding, La France insoumise overtook the Socialists, Marine Le Pen reached the 2017 runoff and her party's list came first at the 2019 European election.
Finding: Engagement on French party pages did not track the vote evenly; the gaps were largest for small, highly engaged parties.
AbstractPDFSummaryFacebook added emoji reactions, including Angry, in February 2016, and for a time weighted them more heavily than likes in ranking posts. Did audiences of European parties become angrier?
Finding: The rise of anger on party pages was mostly a rise in emoji use.
AbstractPDFSummaryOn Facebook, a presidential campaign is carried by more pages than the candidate's own. I compare 17 pages on the two sides of the 2020 United States presidential campaign: the candidates', running mate's, campaign and party pages, several Trump family pages, and a campaign-linked page and a handful of supporters' pages on Biden's side.
AbstractPDFSummaryFacebook is often said to reward the American right. I test the claim on matched pairs: the Facebook pages of the Democratic and Republican nominees in 20 contested 2020 US Senate races, over the final eight weeks before 28 October 2020 (8,125 posts from CrowdTangle). Democrats posted about twice as often as their opponents (median 30 against 19 posts a week).
AbstractPDFSummaryDo politicians set the tone of the conversations beneath their posts?
AbstractPDFSummaryIn 2016 the comment sections under Donald Trump's and Hillary Clinton's Facebook posts drew millions of comments. Who took part, how hostile were they, and was hostility rewarded? I study 5.2 million comments: every comment that could be retrieved from 730 posts on Trump's page during the Republican primaries (January to April 2016), and from 16 posts on the two candidates' pages on 1 October 2016.
AbstractPDFSummaryDo the political videos people watch online tell us how their area will vote? I test this with a record, by county and by day, of YouTube views of Donald Trump and Hillary Clinton content from 1 January to 15 October 2016, covering 2,184 US counties in 49 states. Counties where Trump took a larger share of candidate views voted more Republican, but nearly all of that is geography and past voting.
AbstractPDFSummaryThis thesis examines the use of Facebook as a campaign tool by British political parties over the last decade. The thesis measures and conceptualises party approaches to Facebook, through examining extracted Facebook posts (n=49,551), hand-coded Facebook posts (code n=882, posts n=1481), Facebook Audience Insights and survey data.
AbstractPDFSummaryWhat do parties actually post, who do they reach, and what do platforms really change? Six ideas from my research. Select a card for the detail, plus links to the papers, posts and videos behind it.
New here? Start with these three
Facebook mostly recreates the traditional campaign, only louder.
One party, different faces for different audiences.
Behind the curtain of the targeting machine.
Adventurous campaigning without risking the brand.
A hidden army of followers, far larger than formal membership.
You can't scrutinise what you can't see.
Explainers, talks and media appearances from my YouTube channel, on campaigns, platforms and the data behind them. Pick a video below to play it here.
If the player does not load, watch on YouTube ↗
An ESRC-funded doctorate on British parties' Facebook campaigns, after an LLB, an MA in Politics and an MRes.
Research consultant during the General Election, analysing targeted political ads using crowdsourced data from volunteers.