Political Effects of the Internet and Social Media — Comprehensive Notes
The Internet, Social Media, and Politics – Comprehensive Notes
Focus of the review: empirical evidence on how Internet and social media influence political outcomes, with emphasis on features that distinguish online from offline media (low barriers to entry, user-generated content) and how these affect voting, protests, polarization, xenophobia, fake news, and autocratic regime strategies.
Key distinctions: online media can mirror traditional media but differ in gatekeeping, speed, reach, interactivity, data availability, and manipulation risks; these shape political consequences.
Overall takeaway: effects are mixed and context-dependent. The Internet and social media can enhance accountability and mobilization in some settings (notably when corruption and autocratic control are prevalent) but can also amplify misinformation, polarization, and manipulation in others. Autocracies increasingly rely on censorship, propaganda, and surveillance online; democracies face new challenges from populism and misinformation, with regulation and policy still unsettled.
Core mechanisms discussed: information access and diffusion, coordination and mobilization, exposure to diverse vs. homogeneous content, social pressure and network structure, and strategic behavior by both citizens and politicians.
Common research challenges: identifying causal effects amid endogenous platform growth, user adoption, and network formation; isolating content effects from network and scale effects; accounting for heterogeneity across countries, institutions, and platforms.
Structure of notes: effects on voting, protests, polarization, xenophobia, false news, politicians’ behavior, autocratic strategies, and open questions for future research.
2. THE INTERNET, SOCIAL MEDIA, AND VOTING
Research design: treat the Internet as a black box to compare Internet access vs. not, and separately study social media content and network effects when possible; many studies rely on natural experiments and instrumental variables to identify causal effects.
General pattern in mature democracies (Internet and voting): early Internet access reduced political interest and turnout via substitution toward online entertainment and reduced demand for political news; later, with social media, new populist actors mobilized through direct online connectivity.
Major empirical findings on the Internet’s effect on voting
Falck, Gold, Heblich (2014): broadband expansion in Germany (2004–2008). Identification using preexisting telephone infrastructure; finding: Internet reduced turnout but did not change party vote shares. Mechanism: substitution of political news with entertainment.
Notation/idea: local variation in Internet deployment affects turnout; no party-level effects detected.
Gavazza, Nardotto, Valletti (2019): UK (2006–2010). Broadband exposure reduced electoral participation; incumbents benefited from the crowding out of political news by online entertainment. Identification via weather shocks causing outages (outage-based instrument).
Campante, Durante, Sobbrio (2018): Italy (1996–2013). Negative effect of Internet on political participation persisted until 2008, after which it reversed with the rise of social networks and online political activity. Five Star Movement gains associated with higher broadband access.
Schaub & Morisi (2019); Guriev, Morisi, Zhuravskaya (2020): Europe-wide evidence that broadband/mobile Internet expansion boosted anti-establishment/populist support (Italy 2013; Germany 2017; broader European data with 3G coverage 2007–2018). Key finding: incumbents lose support with greater Internet expansion; populist opposition parties gain.
Miner (2015): Malaysia elections (2004 & 2008). Broadband growth led to substantial decline in incumbent support in connected areas; associated with regime turnover potential due to information access in previously censored contexts.
Donati (2019): South Africa (2006–2016). 3G access reduced ruling party vote share; effects stronger in more corrupt localities; mobile Internet facilitates political participation and electoral competition.
Guriev et al. (2020): Gallup World Poll, 116 countries (2008–2017). 3G expansion increased perceived corruption and reduced government approval where Internet is not censored; stronger effects where traditional media are censored. Regions with no 3G penetration show no correlation between actual corruption and perceived corruption.
Key implications
In censored or information-poor environments, Internet access substantially broadens political information, potentially facilitating regime change or accountability.
In mature democracies, the Internet’s net effect on participation is nuanced: entertainment/low-news exposure can depress turnout in the short run; political information online and social media can alter mobilization dynamics, sometimes benefiting anti-establishment movements.
The Internet vs. social media distinction in voting
Bond et al. (2012): Facebook field experiment during the 2010 US congressional elections (61 million participants). Treatment with a political-information message plus social cues (neighboring friends who reported voting) increased self-reported turnout; effect attenuated when turnout was validated; stronger effects with higher user-to-user interaction intensity.
Jones et al. (2017): replication in the 2012 US election; similar patterns of social influence on voting.
Rotesi (2018): Twitter penetration and turnout in 2008/2012 US elections; instrumental variation via shocks to regional Twitter usage from sports-team player trades; results: instrumented Twitter penetration correlates with higher turnout and lower Democratic vote share (Obama), suggesting social influence/peer effects; however, interpretation is local and may reflect politics of sports-followers rather than broad electorate.
Overall synthesis on voting and the Internet
The Internet and social media have contributed to populist success in parts of Europe and to reduced support for incumbents in some developing/semi-authoritarian contexts.
Social media can mobilize voters, but effects are heterogeneous across contexts, platforms, and subpopulations.
3. THE INTERNET, SOCIAL MEDIA, AND PROTESTS
Core distinction: protests in autocratic regimes often rely on street action (when elections are non-credible) and social media can reduce coordination costs and information frictions, increasing participation.
Mechanisms by which social media can influence protests
Low barriers to entry enable dissemination of anti-government information and reduce the cost of mobilization.
Horizontal information flows help coordinate timing and tactics (logistics, locations, shifts).
Content and network structure can reduce coordination costs and facilitate rapid mobilization, especially in repressive environments where traditional media are controlled.
Evidence linking social media to off-line protests
Sina Weibo data (Qin, Strömberg, Wu, 2017): 13.2 billion posts (2009–2013) with protest/strike keywords predict off-line events a day ahead; high information content on local corruption correlates with future protests.
Acemoglu, Khandelwal, Kuran? (2018): Twitter signals Tahrir Square demonstrations; results consistent with social media aiding mobilization during Egypt’s Arab Spring.
Steinert-Threlkeld, Mocanu, Vespignani, Fowler (2015): 14 million geolocated tweets across 16 countries; coordinated hashtag campaigns linked to more protests the following day during Arab Spring; evidence of online coordination translating into offline action.
Hendel, Nyhan, Reifler (2017): cottage cheese boycott in Israel (2011) coordinated on Facebook; strong sales declines with high Facebook penetration; shows online coordination affecting private-sector outcomes and real-world behavior.
Amomir/Amorim et al. (2018): Occupy movement in the United States; addition of a local ISP in a locality (roughly +0.5 percentage points in penetration) increases the probability of Occupy protests by about 1–3 percentage points.
2G/3G and Africa: Manacorda & Tesei (2020) show that broader mobile network expansion (2G GSM) across Africa increased anti-government protests, especially during economic downturns; identified using a weather-based/shoc-based instrument (lightning strikes affecting adoption).
Causality and identification challenges
Establishing causal effects of social media on protests is difficult due to endogeneity in content, networks, and the timing of protests.
Studies with strong identification (e.g., VK in Russia, 2011–2012 protests) show that social media penetration can raise protest participation via coordination costs and social influence rather than purely information content.
Cross-country and contextual heterogeneity
Amorim et al. show Occupy was partially facilitated by Internet access, but effects depend on offline grievances and access to online platforms.
China-related studies emphasize censorship and selective visibility: social media can coordinate local information flows (local protests) while censorship suppresses national mobilization signals.
Key takeaway on protests
Convincing evidence that low entry barriers and horizontal information flows via social media facilitate protests, particularly where political information is scarce or censored and where coordination can overcome collective-action problems.
Causal effects vary by context, network structure, and the presence of alternative offline coordination channels.
4. THE INTERNET, SOCIAL MEDIA, AND POLARIZATION
Central debate: do online networks create echo chambers and increase political polarization, or do they expose people to diverse views and reduce polarization?
Evidence on exposure to ideologically diverse content
Bakshy, Messing, Adamic (2015): on Facebook, users encounter less content aligned with opposing ideologies due to algorithmic feeds and selective sharing; however, user homophily (liking similar people) is a major driver of exposure to like-minded content.
Halberstam & Knight (2016): Twitter networks among politically engaged users show strong partisan clustering in retweets; mentions of opposing camps are more common than linking to them, but a small heterogenous cluster exists where cross-ideology interaction occurs.
Conover et al. (2011): retweet networks are highly polarized; interaction with opposing camps is limited; however, cross-cutting mentions occur, suggesting partial exposure to opposing views exists.
Gentzkow & Shapiro (2011): online vs offline exposure to news: online segregation in news consumption is similar to offline, but online social interactions are less segregated than offline media consumption.
Polarization in online vs offline interactions
Sunstein’s echo-chamber hypothesis has driven much of the literature; mixed empirical support.
Barberá (2015): social media can reduce mass polarization by expanding exposure to weak ties; diverse networks associated with moderation; argues that online exposure to weak ties may lower polarization.
Mixed recent evidence on polarization dynamics
Boxell, Gentzkow, Shapiro (2017): polarization trends in the US show increases among groups less likely to use the Internet; suggests Internet use is not the sole driver of rising polarization.
Lelkes, Sood, Iyengar (2017): broadband access increases partisan hostility and the consumption of partisan media, implying potential polarization channels; does not establish a direct causal link from Internet to polarization.
Yanagizawa-Drott, Rao, Petrova, Enikolopov, Bursztyn (2020): Facebook network homophily correlates with greater voting homophily; online homophily may feed polarization.
Allcott et al. (2020) & Mosquera et al. (2020): field experiments deactivating Facebook show reductions in political polarization (Allcott et al.); Mosquera et al. find reduced ability to recognize bias after temporary deletion but not a direct polarization measure.
Net interpretation
The literature finds no clear, universal verdict: online exposure can both mitigate and amplify polarization depending on network structure, user behavior, platform design, and prevailing political incentives.
The social value of connectivity (consumer surplus) is robust, even when political polarization effects are ambiguous.
5. SOCIAL MEDIA AND XENOPHOBIA
Hypothesized channels: social media could (a) coordinate hate crime via collective action, (b) alter attitudes by exposure to xenophobic content, (c) reduce stigma by providing a platform for hate-driven dialogue.
Evidence on hate crimes and attitudes
Bursztyn, Egorov, Enikolopov, Petrova (2019): Russia (2007–2015). Using the Enikolopov et al. (2020) approach, social media penetration (VK) increased hate crimes in cities with high baseline nationalist sentiment, but not in low-nationalism cities. Long-run cross-sectional evidence; mechanism includes attitude shifts captured via a list experiment showing persuasive effects especially among young and less educated; stronger for organized crimes by multiple perpetrators (coordination) than for solo acts.
Müller & Schwarz (2018, 2019): Germany (2015–2017). Anti-refugee sentiment on AfD Facebook pages correlates with higher hate crimes against refugees in places with higher social media usage; the effect weakens when there are outages or competing news cycles.
Additional lines of evidence point to high-profile dissemination of hateful messages by opinion leaders on social media and its link to hate crimes, though identification strategies vary.
Mechanisms and interpretation
Coordination: social media reduces costs and frictions for coordinating hate crimes and collective actions.
Attitude formation: exposure to xenophobic messaging can shift attitudes toward more extreme positions, especially for vulnerable groups (young, less educated).
Stigmatization: social media may lower social stigma associated with xenophobic attitudes, facilitating action.
Overall takeaway on xenophobia
There is credible evidence that social media exposure can increase xenophobic attitudes and, in certain contexts, hate crimes, particularly where nationalist sentiment is already present and coordination is feasible.
6. SOCIAL MEDIA AND FALSE NEWS
Core concern: rapid spread of false news via social media; debates on prevalence vs. persuasion.
Key empirical findings on the diffusion of false news
Mocanu et al. (2015): Italy (2012 elections). False news spread rapidly on social media, especially among users who mistrust mainstream media.
Allcott & Gentzkow (2017): US 2016 presidential campaign. Identified 115 pro-Trump and 41 pro-Clinton fake stories circulating; 30 million shares for pro-Trump fake news and 8 million for pro-Clinton. Post-election survey: about 15% recalled seeing false news; 8% believed one; exposure to placebo false stories (uncirculated) yielded similar belief rates, highlighting recall limitations. On average, a US voter was exposed to about one or more fake articles.
Guess, Nagler, Tucker (2019); Guess, Lach, Reifler (2018): browsing histories show Facebook markedly more likely to be visited immediately before a false-news site than a hard-news site; Twitter and Google show similar exposure patterns to real vs. false content; false-news sharing varies by partisanship and age.
Grinberg, Joseph, Friedland, Swire-Thompson, Lazer (2019): Twitter shares of false news constitute about 6% of political news; exposure is highly concentrated: ~1% of users account for ~80% of exposures; retrospectives show false-news retweets cluster among extreme right-leaning users.
Vosoughi, Roy, Aral (2018): 2006–2017 Twitter data. False stories diffused faster, more broadly, and deeper than true ones, especially for political content; top percentile of viral false news reached 1,000–100,000 users, while true stories rarely exceeded 1,000.
Henry, Zhuravskaya, Guriev (2020): France (2019 EU Parliament election). A field experiment shows fact-checking reduces willingness to share false statements by about 30%; right-wing respondents more likely to share false news; introducing friction (more clicks) substantially reduces sharing.
Persuasion vs. diffusion
Evidence on the persuasive effect of exposure to false news remains contentious; some experiments show substantial persuasion effects, while others emphasize the role of prior attitudes and limited exposure.
Several studies highlight that social media is a major diffusion channel for false news, with concentration in certain platforms (notably Facebook and Twitter) and demographic groups.
Policy implications and gaps
Fact-checking can reduce sharing propensity but may not fully mitigate belief in false stories; the causal chain from exposure to persuasion is complex and platform-dependent.
The literature calls for more evidence on the direct persuasion effect of online false news, especially within social-media sharing networks, and on how platform design (friction, ranking) shapes dissemination and belief.
7. THE INTERNET, SOCIAL MEDIA, AND POLITICIANS’ BEHAVIOR
How politicians respond to online information environments
Bessone, Campante, Ferraz, Souza (2019): 3G mobile technology access leads to more active Facebook engagement by legislators; online engagement correlates with higher incumbent vote share in subsequent elections; but online activity crowds out offline political activity (fewer speeches; reduced earmarked transfers to connected municipalities), signaling potential accountability trade-offs.
Enikolopov, Makharin, Petrova (2018): blogging about corruption in large state-owned firms associated with management turnover and improved governance in the long run; stock market effects observed.
Theocharis et al. (2016): political conversations on Twitter in the run-up to 2014 European elections show high levels of uncivil behavior and harassment toward politicians; this may push politicians toward one-way broadcasting rather than interactive dialogue.
Petrova, Sen, Yildirim (2020): Twitter presence helps candidates in US Congress raise campaign contributions, especially for newcomers; online presence helps new entrants gain visibility and funding.
Implications for accountability and governance
Online platforms can enhance voter information and accountability by increasing visibility of politicians’ actions; yet, the same environment can reduce offline engagement (e.g., traditional campaigning and in-person connections).
Civil discourse concerns may push political actors toward broadcast-style communication, potentially narrowing deliberation.
Summary
Social media changes in politicians’ behavior and campaign finance dynamics, with a mix of increased online responsiveness and potential reductions in offline political activity; more research needed to draw firm conclusions about broader accountability effects beyond elections and street protests.
8. ONLINE STRATEGIES OF AUTOCRATIC REGIMES
How autocrats use the Internet and social media to sustain control and manage dissent
8.1 Digital Censorship
China as a focal case: selective deletion and censorship as the primary tool, with many posts about collective action being censored more than posts critical of the regime.
King, Pan, Roberts (2013): observational snapshots show censorship of mobilization-related posts more than anti-government content.
King, Pan, Roberts (2014): randomized blog postings reveal that posts about real-world collective action are censored; posts critical of the state are less likely to be censored than those involving mobilization.
Self-censorship and frictions
Roberts (2018): even limited friction in accessing censored information can deter users; partial censorship can obscure the fact that information is censored, reducing demand for circumvention tools.
Mixed effects of censorship
Hobbs & Roberts (2018): attempts to block content (e.g., Instagram) can trigger a surge in VPN adoption; users then access other censored platforms and potentially engage more with politically sensitive topics.
Experiments on uncensored access
Chen & Yang (2019): Beijing students given free VPN access for 18 months; without extra encouragement, many did not actively browse censored content; with incentives to engage with foreign news, consumption of politically sensitive information increased and persisted after incentives ended.
Takeaway on censorship
Digital censorship can be effective but is not foolproof; partial censorship and strategic dissemination can influence information environments in ways that may increase political engagement or alter knowledge formation.
8.2 Manipulation of Information
Governments manipulate online discourse to shape the political agenda
King, Pan, Roberts (2017): analysis of China shows a massive government effort to post pro-state content and engage in agenda-setting and framing rather than meaningful argument.
Foreign interference and bots
Martin et al. (2018): foreign influence campaigns identified across multiple countries; Russia prominent in many cases; bots and disinformation play roles in shaping public opinion.
Gorodnichenko, Pham, Talavera (2018): aggressive use of Twitter bots during the 2016 US elections and Brexit; manipulation via bots could contribute to outcomes in fragmented social media environments.
Counterpoints and caveats
Bail, Makarov, et al. (2020): analysis of Russian Internet Research Agency activity suggests limited short-term impact on political attitudes and social-media behavior in late 2017; argues that trolls might have failed to seed polarization if they mainly engaged with already polarized individuals.
Implications
State and foreign manipulation via social media is plausible and has measurable effects on discourse, but the causal impact on political outcomes is nuanced and context-dependent.
8.3 Monitoring and Surveillance
Social media as a tool for governance and control
Governments leverage online data to monitor public sentiment and to detect protest signals, particularly in autocratic regimes where official transparency is limited.
Chinese authorities’ use of online signals to identify discontent and intervene at local levels is documented in multiple studies (e.g., Egorov et al. 2009; King et al. 2013, 2014; Lorentzen 2014).
Implications
Digital traces create incentives for citizens to optimize online behavior to avoid detection, while authorities gain more granular, real-time surveillance capabilities.
Takeaway
Surveillance and selective information strategies can be effective tools for authoritarian control, complementing formal censorship and propaganda.
9. MAIN LESSONS AND QUESTIONS FOR FUTURE RESEARCH
Broad lessons
In environments where corruption and autocrat-controlled media are prevalent, free Internet and social media tend to improve accountability by informing the public and enabling protest coordination.
Autocrats respond with censorship, selective information suppression, and misinformation to dampen accountability pressures.
In established democracies, social media can amplify grievances contributing to populism, though evidence on broad-based polarization remains mixed; younger cohorts may be more resilient to fake news.
Open questions and future research priorities
Are observed effects temporary as users adapt and become more discerning, or are they persistent across generations? Early evidence suggests younger users may be less affected by fake news, but long-run dynamics require more study.
Do intrinsic properties of social media (format, speed, and brevity) favor extreme, xenophobic, or populist messages, or can platforms and institutions channel these tendencies toward constructive outcomes?
How does the structure of online networks influence political participation and policy preferences? More work is needed on network topology, information diffusion, and coordination with robust causal identification.
Regulatory policy questions: what kinds of regulation best balance free speech with the mitigation of hate speech and misinformation? The evidence to date does not yield a single clear policy solution.
Final synthesis
The Internet and social media are powerful tools that reshape political behavior, with both positive (accountability, mobilization) and negative (misinformation, polarization, hate speech) implications.
The effects are highly contingent on institutional context, platform design, network structure, and the presence of concurrent political and economic grievances.
Continued rigorous empirical work with credible identification strategies is essential to disentangle mechanisms and inform policy.
References to illustrative studies (selected examples)
Falck, Gold, Heblich (2014); Gavazza, Nardotto, Valletti (2019); Campante, Durante, Sobbrio (2018)
Miner (2015); Donati (2019); Guriev, Morisi, Zhuravskaya (2020)
Bond et al. (2012); Jones et al. (2017); Rotesi (2018)
Qin, Strömberg, Wu (2017, 2019); Acemoglu et al. (2018); Steinert-Threlkeld et al. (2015)
Barberá (2015); Boxell, Gentzkow, Shapiro (2017); Lelkes, Sood, Iyengar (2017); Yanagizawa-Drott et al. (2020); Allcott et al. (2020); Mosquera et al. (2020)
Bursztyn, Egorov, Enikolopov, Petrova (2019); Müller & Schwarz (2018, 2019); Henry, Zhuravskaya, Guriev (2020)
Enikolopov et al. (2020); King, Pan, Roberts (2013, 2014, 2017); Roberts (2018)
Martin et al. (2018); Gorodnichenko et al. (2018); Bail et al. (2020)
Petrova, Sen, Yildirim (2020)
Note: The above is a curated set of cited works to illustrate the points in this summary. The original article includes a comprehensive bibliography.
{ "title": "Political Effects of the Internet and Social Media — Comprehensive Notes", "note": "" }