The arc of the field: from supply-side observation to causal effects under platform regulation
Research on elections and social media has, over roughly a decade, traced an arc from descriptive mapping of partisan attention through accounts of coordinated manipulation toward causally identified effects of platforms and political advertising — and, most recently, toward a sober reckoning with what kinds of knowledge are even possible under shifting platform access. Italian work anchors the earlier portion of this arc: Iannelli2015-e0818c3e situates Twitter as a “second screen” hybridised with TV talk shows during the 2013 campaign, while Giglietto2019-882f1900 uses the 2018 vote to show that populist (especially M5S) news sources are more insular on Twitter and that insularity correlates with amplification-by-sharing rather than contestation-by-commenting on Facebook. F2020-6278a4aa reads these dynamics through three lenses — networked publics, micro-targeted ads, and attention competition — that recur as organising motifs in much of what follows.
Coordination, hyperpartisanship, and the brokerage of manipulation
A substantial line of inquiry treats elections as sites of coordinated and inauthentic behaviour. Giglietto2020-9d8acdd7 introduces “coordinated link sharing behaviour” during Italy’s 2018 and 2019 elections, arguing for an ecological, action-based reframing of disinformation research; Giglietto2023-fa71a001 operationalises this insight as a continuous detection workflow, surfacing hyperpartisan M5S networks alongside religious clickbait operations. Oprea2025-lf extends the same logic to Romania’s 2024 EP campaign, while Kansaon2025-id migrates coordination analysis into WhatsApp’s closed Brazilian ecosystem, connecting message-level coordination to street-level mobilisation against the Supreme Court. Gaw2025-ru reconceptualises such activity as political brokerage in the 2022 Philippine elections, foregrounding the political-economic infrastructures that sustain it. At the global scale, Appel2026-qr and its companion noauthor_undated-bm measure the reach of 49 Meta-identified deceptive networks around the 2020 US election, finding that exposure is highly concentrated, that financially-motivated operations engage heavily in political content, and that reshares by non-network users — not network accounts themselves — drive most exposure. Knupfer2025-vt complements these findings with a theoretical move: a “logic of connective faction” in which digitally networked elites and hyperpartisan media co-produce intra-party radicalisation, illustrated by the CRT issue.
Misinformation: from elite supply to citizen demand and downstream attitudes
A parallel cluster focuses on who spreads misinformation, who believes it, and with what consequences. Tornberg2025-ir establishes, across 26 countries, that neither populism nor right-wing ideology alone predicts elite misinformation sharing — only their interaction does — naming radical-right populism as the central driver. Tai2026-qk extends this to US elected officials, while Gonzalez-Bailon2024-rq quantifies the diffusion patterns that give misinformation its distinctive reach on Facebook. On the demand side, Rossini2026-jn uses a Brazilian three-wave panel to show that belief in electoral misinformation increases political intolerance and that messaging-app news use raises intolerance indirectly through misinformation beliefs; Ventura2025-sw complements this with a WhatsApp deactivation experiment. Van_Erkel2026-mk introduces a “hostile misinformation effect” — citizens think their own side is the target — which mirrors the older hostile media effect and erodes trust in electoral integrity. Gattermann2025-yx links such concerns to far-right performance at the 2024 EP elections. Two papers from the Center for an Informed Public (Starbird2025-jj, Prochaska2025-ef) reframe election misinformation as collective sensemaking, arguing that the locus of falsehood lies in framings layered onto often-accurate evidence and in “deep stories” that audiences activate from elite cues.
Visibility, virality, and the affective economy of campaigns
A further strand examines how partisan supply and platform affordances jointly shape attention. Larsson2026-ro tracks a decade of Norway’s Progress Party on Facebook and shows that negative posts drive shares and comments while positive posts drive likes — a finding echoed cross-nationally by Kalsnes2025-zb on Scandinavian “Angry” and “Love” reactions, and surveyed comprehensively by Anwar2024-34dba628. Iris2026-pg documents disproportionate radical-right visibility across five EU countries’ online news during the 2024 EP campaign, while Bouchafra2026-ts zooms in on the Sweden Democrats’ platform-tailored visual securitisation across Facebook, X and TikTok. Achmann-Denkler2026-lx and Schulte2026-df push the visual and multimodal agenda methodologically into German campaigns, the latter explicitly framing multi-platform observation as a research-infrastructure problem. Balluff2026-bv shows how government advertising itself can capture media visibility in Austria’s “Inseratenaffäre,” while Rodarte2026-dk documents Brazilian parliamentarians’ contests for epistemic authority over journalistic gatekeeping during the Manaus collapse. Arceneaux2026-xk adds a non-human actor to this picture, arguing that social bots function as agenda-builders especially for attribute (sentiment) salience.
Algorithms, advertising, and the search for causal effects
The boldest recent move is to causally identify what platforms — not just users or campaigns — do to electoral outcomes. Gauthier2026-iq reports a field experiment showing that X’s “For You” algorithm shifts attitudes rightward asymmetrically and persistently, because it induces users to follow conservative activist accounts who remain in their feeds afterwards. Efstratiou2025-gs audits the same platform from a different angle, attributing right-leaning visibility gains less to political bias per se than to agitation and proximity to Elon Musk. On political advertising, Kim2026-wg documents geo-racial targeting and turnout depression by 2016 Russian voter-suppression ads — a sharply non-null result obtained by measuring direct exposure and verified turnout, in contrast to Allcott2025-jb’s null effects from Meta’s 2020 ad-removal experiments. Votta2025-xz adds the political economy of ad delivery itself, while Inacio-da-Silva2026-zf documents independent crowd-sourced auditing of Brazilian Facebook ads. Lin2025-xp opens a frontier with AI-driven persuasion experiments across four elections, finding dialogue effects larger than traditional ads and curiously asymmetric in factual accuracy across the ideological spectrum. Giglietto2025-1765bb4f inverts the question: rather than asking what platforms do to politics, it documents how Meta’s undisclosed political-content reduction policy cut Italian MPs’ Facebook reach by 72%, with extremist accounts compensating through sheer posting volume.
Reflexivity: what can social science actually know?
This empirical maturation has prompted a sharp metascientific turn. Munger2025-cz argues that even the celebrated Meta2020 partnership reveals the structural inadequacy of empirical social science for governing rapidly mutating platforms, calling for proactive regulation that obliges companies to characterise their own products. Philipp2026-tl, Lukito2026-nb, and Jurg2025-ur develop this concern into a programme: data access is now structured by price, proficiency, and permission, with DSA-mandated APIs offering uneven and ephemeral relief, and platform content-moderation transparency itself in need of auditing. Schulte2026-df and Giglietto2025-1765bb4f further argue that observability of elections must be conceptualised as reusable research infrastructure rather than as one-off study design. Taken together, these papers refocus the field’s central question: not just what happens in elections on social media, but what the conditions of possibility are for knowing it — a question whose answer increasingly depends on regulation, comparative cross-platform design, and a willingness to study both supply and demand sides of political communication in the same breath.