Facebook reactions in the context of politics and social issues: a systematic literature review

Summary

This systematic literature review synthesizes 64 peer-reviewed articles (2016–2023) on Facebook’s “Reactions” feature (Love, Haha, Wow, Sad, Angry), introduced in 2016 as an expansion of the Like button. The authors organize the field into three thematic clusters — the introduction of Reactions, politics and far-right groups, and other social issues — and perform a meta-analysis across them. The central argument is that Reactions function as paralinguistic digital affordances that capture nuanced affect with low cognitive cost, and that they reveal a systematic asymmetry: lifestyle and entertainment content draws predominantly positive reactions, while sociopolitical content — especially from populist and far-right actors — attracts broader and more negative emotional spectra, with anger consistently driving higher engagement.

Key Contributions

• First comprehensive systematic synthesis of Facebook Reactions research from 2016 through 2023, spanning political communication, health, business, and social-issue domains.
• A thematic taxonomy that organizes a fragmented, interdisciplinary literature into coherent categories.
• Identification of cross-study regularities, especially the role of negative emotion (anger in particular) in amplifying engagement with political and populist content.
• Diagnosis of methodological gaps: lack of unified data sources, limited cross-cultural comparison, and underexplored event-based content.
• Suggestions for future directions, including LLM-based categorization of reaction patterns and applications to misinformation, fact-checking, and digital activism.

Methods

The authors follow Briner and Denyer’s five-step systematic review process: formulating research questions, searching, selecting and evaluating studies, analyzing, and reporting. Database queries in Scopus, Google Scholar, and other engines (July–September 2023) using Boolean combinations of “Facebook” and “Reaction” yielded 64 peer-reviewed journal articles and conference proceedings. Each was read and coded using Armstrong’s comparative naturalistic method, then grouped into three top-level categories with finer sub-topics (political news, populism, healthcare, business, emotion detection).

Findings

• Of 64 articles, 6 addressed the introduction of Reactions, 19 covered politics and far-right groups, and 39 examined other social issues.
• In political campaigns (Mexico, Brexit, EU Parliament), anger-laden appeals produced significantly more engagement and sharing than positive content.
• Populist and far-right leaders disproportionately attract Angry and Haha reactions relative to mainstream politicians, with anger boosting sharing.
• Hyperpartisan and junk news exhibit a bivalent emotional logic: Love and Angry rarely co-occur, while Angry tends to co-occur with Sad and Wow.
• Reactions have been used computationally to predict emotion classes, to measure controversy via entropy of reaction distributions, and as sentiment labels — including for low-resource languages like Sinhala and Bangla.
• Love and Haha generally signal positivity; Sad and Angry are reliably negative; but Haha and Wow can be ambiguous or hostile in political contexts (e.g., Wow as disbelief).
• Lifestyle and entertainment posts skew positive; sociopolitical posts elicit broader, more negative emotional profiles.
• Cultural context shapes reaction patterns, as in studies contrasting China-critical, China-supporting, and neutral media audiences in Hong Kong.

Connections

This review consolidates a research tradition that several papers in the same register extend empirically: work using Reactions and similar engagement signals to detect coordinated or hyperpartisan amplification connects directly to Giglietto2022-b30e8b4e and Giglietto2024-cbeb3f70, while the focus on anger and negative affect driving political engagement resonates with studies of partisan and far-right discourse such as Ghezzi2023-8bebc91f and Bruns2023-039725ce. The methodological commentary on using lightweight platform affordances as proxies for sentiment and controversy also relates to broader algorithmic-exposure findings in Bakshy2015-rn.