Matteo CinellipersName, Gianmarco De Francisci MoralespersName, Alessandro GaleazzipersName, Walter QuattrociocchipersName, Michele StarninipersName (2021date). “The echo chamber effect on social media.” Proceedings of the National Academy of Sciencestitle·venue.
Object of study — echo chambersterm on social media - whether they exist, and whether their existence is a property of online interaction or of a particular platform's design
Research question — whether the echo chamberterm effect survives a comparison across platforms with different interaction paradigms, measured the same way on each
Contribution — an operational definitionterm·mth of echo chambersterm - individual leaning plus interaction structure, quantified independently - applied identically to Facebookterm, Twitter, Redditterm and Gabterm over 100 million pieces of content
phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | aggregation of users into homophilic clusters, and the bias it puts on which information reaches whom - present on some platforms and absent on others — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | |
| claim-type | existence — “These results indicate that information diffusion is biased toward individuals who share a similar leaning in some social media, namely Twitter and Facebook. In contrast, in others—Reddit and Gab in our analysis—this effect is absent.” | |
| method — evidence | — “Facebook and Reddit are particularly apt to a cross-comparison since they share the definition of individual leaning (computed by using the classification provided by mediabiasfactcheck.org; see Materials and Methods for further details) and the rationale in creating connections among users that is based on an interaction network.” | |
| object of study — evidence | — “We can broadly define echo chambers as environments in which the opinion, political leaning, or belief of users about a topic gets reinforced due to repeated interactions with peers or sources having similar tendencies and attitudes.” | |
| research question — evidence | — “We introduce an operational definition of echo chambers to provide a common methodological ground to explore how different platforms influence their formation. In particular, we operationalize the two common elements that characterize echo chambers into observables that can be quantified and empirically measured, namely, 1) the inference of the user’s leaning for a specific topic (e.g., politics, vaccines) and 2) the structure of their social interactions on the platform.” | |
| contribution — evidence | — “We introduce an operational definition of echo chambers to provide a common methodological ground to explore how different platforms influence their formation. In particular, we operationalize the two common elements that characterize echo chambers into observables that can be quantified and empirically measured, namely, 1) the inference of the user’s leaning for a specific topic (e.g., politics, vaccines) and 2) the structure of their social interactions on the platform.” | |
| transfer | from epidemiology: the susceptible-infected-recovered epidemic model, carried from epidemiology to measure which users a message reaches and with what leaning — “These results indicate that information diffusion is biased toward individuals who share a similar leaning in some social media, namely Twitter and Facebook. In contrast, in others—Reddit and Gab in our analysis—this effect is absent.” | transfers[0] |
| transfer | from network science: homophily and community detection, carried from network science into a test of a psychological construct — “Then, we use these elements to assess echo chambers’ presence by looking at two different aspects: 1) homophily in interactions concerning a specific topic and 2) bias in information diffusion from like-minded sources.” | transfers[1] |
| mechanism | homophilic clustering biasing diffusion when a user's neighbours share their leaning, a message released by that user reaches an audience with the same leaning; running an epidemic model over the interaction network turns that structural fact into a measurable dependence of the influence set's leaning on the seed's — “These results indicate that information diffusion is biased toward individuals who share a similar leaning in some social media, namely Twitter and Facebook. In contrast, in others—Reddit and Gab in our analysis—this effect is absent.” | mechanisms[0] |
| mechanism | feed algorithm as the discriminator platforms that choose a user's content for them, and platforms where the user selects their own communities, produce different network topologies from the same human tendencies - so the effect divides along a design decision rather than along the population — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | mechanisms[1] |
| mechanism | saturation hiding as absence a measurement of contrast between leanings can only see a difference that exists inside the sample; if a platform's whole population shares one leaning, the measurement returns the same null it would return if there were no echo chamber at all — “Such a latter configuration may depend upon two factors: 1) Gab and Reddit are not bursting the echo chamber effects, or 2) we are observing the dynamic inside a single echo chamber.” | mechanisms[2] |
| qualifies | single-platform findings about echo chambers, which cannot distinguish a property of online interaction from a property of one platform's design — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | claim_relations[0] |
| builds-on | group polarization theory and the selective-exposure account of why echo chambers form — “According to group polarization theory, an echo chamber can act as a mechanism to reinforce an existing opinion within a group and, as a result, move the entire group toward more extreme positions.” | claim_relations[1] |
| reconciles | the dispute over whether echo chambers exist at all - by showing the answer differs by platform under one measurement — “Echo chambers have been shown to exist in various forms of online media such as blogs, forums, and social media sites. Some studies point out echo chambers as an emerging effect of human tendencies, such as selective exposure, contagion, and group polarization. However, recently, the effects and the very existence of echo chambers have been questioned. This issue is also fueled by the scarcity of comparative studies on social media, especially concerning news consumption.” | claim_relations[2] |
| sense | echo chamber effect - ON FACEBOOK AND TWITTER: strong correlation between a user's leaning and their neighbourhood's, communities spanning the spectrum but internally homogeneous, and diffusion biased toward like-minded peers — “These results indicate that information diffusion is biased toward individuals who share a similar leaning in some social media, namely Twitter and Facebook. In contrast, in others—Reddit and Gab in our analysis—this effect is absent.” | senses[0] |
| sense | echo chamber effect - ON REDDIT AND GAB: absent. Users form a single community rather than opposed clusters, and the leaning of the influence set does not depend on the leaning of the seed — “The corresponding plots in Fig. 1 display a single bright area, indicating that users do not split into groups with opposite leaning but form a single community, biased to the left (Reddit) or the right (Gab).” | senses[1] |
| sense | that absence - READING 1: the platform is not producing echo chambers, which is what the measurement appears to say — “Such a latter configuration may depend upon two factors: 1) Gab and Reddit are not bursting the echo chamber effects, or 2) we are observing the dynamic inside a single echo chamber.” | senses[2] |
| sense | that absence - READING 2: the whole platform IS a single echo chamber, so there is no contrast left inside the data to detect. The authors name both and do not choose — “Such a latter configuration may depend upon two factors: 1) Gab and Reddit are not bursting the echo chamber effects, or 2) we are observing the dynamic inside a single echo chamber.” | senses[3] |
| control | comparing Facebook and Reddit on one common topic - news consumption - because the cross-platform differences could otherwise be attributed to the different topics — held — “One can see that all three measures confirm the picture obtained for other datasets: On Facebook, we observe a clear separation among users depending on their leaning, while, on Reddit, users’ leanings are more homogeneous and show only one peak.” — rules out a cross-platform difference that is a difference of topic, since the platforms are compared on one common topic — “The striking differences observed across social media, in terms of homophily in the interaction networks and information diffusion, could be attributed to the different topics taken into account. For this reason, here we compare Facebook and Reddit on a common topic, news consumption.” | controls[0] |
| control | requiring polarization and homophilic interaction to be quantified independently, so that neither is evidence for the other — held — “Then, we use these elements to assess echo chambers’ presence by looking at two different aspects: 1) homophily in interactions concerning a specific topic and 2) bias in information diffusion from like-minded sources.” — rules out polarization and homophily each being taken as evidence for the other, which would make an echo chamber unfalsifiable — “Crucially, the two elements characterizing the presence of echo chambers, polarization and homophilic interactions, should be quantified independently.” | controls[1] |
| control | three separate measures applied to the same comparison - neighbour-leaning correlation, community structure, and SIR influence sets — held — “One can see that all three measures confirm the picture obtained for other datasets: On Facebook, we observe a clear separation among users depending on their leaning, while, on Reddit, users’ leanings are more homogeneous and show only one peak.” — rules out an echo-chamber result that exists only under one operationalisation of clustering — “One can see that all three measures confirm the picture obtained for other datasets: On Facebook, we observe a clear separation among users depending on their leaning, while, on Reddit, users’ leanings are more homogeneous and show only one peak.” | controls[2] |
| control | choosing two platforms that already share a leaning definition and a network rationale, so the cross-comparison is not itself a methodological artefact — held — rules out that a difference between Facebook and Reddit is a difference in how leaning was defined or how connections were counted, rather than in the platforms — “Facebook and Reddit are particularly apt to a cross-comparison since they share the definition of individual leaning (computed by using the classification provided by mediabiasfactcheck.org; see Materials and Methods for further details) and the rationale in creating connections among users that is based on an interaction network.” | controls[3] |
| weight | single-platform findings about echo chambers, which cannot distinguish a property of online interaction from a property of one platform's design — 0.9 — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | claim_relations[0] |
| weight | group polarization theory and the selective-exposure account of why echo chambers form — 0.85 — “According to group polarization theory, an echo chamber can act as a mechanism to reinforce an existing opinion within a group and, as a result, move the entire group toward more extreme positions.” | claim_relations[1] |
| weight | the dispute over whether echo chambers exist at all - by showing the answer differs by platform under one measurement — 0.85 — “Echo chambers have been shown to exist in various forms of online media such as blogs, forums, and social media sites. Some studies point out echo chambers as an emerging effect of human tendencies, such as selective exposure, contagion, and group polarization. However, recently, the effects and the very existence of echo chambers have been questioned. This issue is also fueled by the scarcity of comparative studies on social media, especially concerning news consumption.” | claim_relations[2] |
| relation | echo chambers —operationalised-as→ observables — the two characterising elements are turned into quantities that can be measured — “We introduce an operational definition of echo chambers to provide a common methodological ground to explore how different platforms influence their formation. In particular, we operationalize the two common elements that characterize echo chambers into observables that can be quantified and empirically measured, namely, 1) the inference of the user’s leaning for a specific topic (e.g., politics, vaccines) and 2) the structure of their social interactions on the platform.” — “we operationalize the two common elements that characterize echo chambers into observables that can be quantified and empirically measured” | concept_edges[0] |
| weight | echo chambers —operationalised-as→ observables — 0.95 — “We introduce an operational definition of echo chambers to provide a common methodological ground to explore how different platforms influence their formation. In particular, we operationalize the two common elements that characterize echo chambers into observables that can be quantified and empirically measured, namely, 1) the inference of the user’s leaning for a specific topic (e.g., politics, vaccines) and 2) the structure of their social interactions on the platform.” | concept_edges[0] |
| differs in (stated by the tagger, no quotation) | Reddit —distinct-from→ Facebook — whether the feed algorithm is tweakable by the user, which is where the echo chamber effect divides | concept_edges[1] |
| relation | Reddit —distinct-from→ Facebook — a user-tweakable feed algorithm against one the user cannot change — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” — “A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | concept_edges[1] |
| weight | Reddit —distinct-from→ Facebook — 0.9 — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | concept_edges[1] |
| outcome | control — held — “The striking differences observed across social media, in terms of homophily in the interaction networks and information diffusion, could be attributed to the different topics taken into account. For this reason, here we compare Facebook and Reddit on a common topic, news consumption.” — “One can see that all three measures confirm the picture obtained for other datasets: On Facebook, we observe a clear separation among users depending on their leaning, while, on Reddit, users’ leanings are more homogeneous and show only one peak.” | controls[0] |
| outcome | control — held — “Crucially, the two elements characterizing the presence of echo chambers, polarization and homophilic interactions, should be quantified independently.” — “Then, we use these elements to assess echo chambers’ presence by looking at two different aspects: 1) homophily in interactions concerning a specific topic and 2) bias in information diffusion from like-minded sources.” | controls[1] |
| outcome | control — held — “One can see that all three measures confirm the picture obtained for other datasets: On Facebook, we observe a clear separation among users depending on their leaning, while, on Reddit, users’ leanings are more homogeneous and show only one peak.” | controls[2] |
| outcome (stated by the tagger, no quotation) | control — held | controls[3] |
| limitation | the null result on Reddit and Gab is explicitly ambiguous - either those platforms are not producing echo chambers, or the observation is being made inside a single one, and the data cannot separate the two — “Such a latter configuration may depend upon two factors: 1) Gab and Reddit are not bursting the echo chamber effects, or 2) we are observing the dynamic inside a single echo chamber.” | limitations[0] |
| limitation | every individual leaning in the study is derived from one external labelling of 2,190 news outlets, so the whole comparison inherits that organisation's judgements — “The labeling of news outlets is based on the information reported by Media Bias/Fact Check (MBFC) (https://mediabiasfactcheck.com), an independent fact-checking organization that rates news outlets on the basis of the reliability and of the political bias of the contents they produce and share. The labeling provided by MBFC, retrieved in June 2019, ranges from Extreme Left to Extreme Right for political bias. The total number of media outlets for which we have a political label is 2,190.” | limitations[1] |
| limitation | two of the Facebook datasets are labelled binary - provaccines/antivaccines, proscience/conspiracy - by the page a post appeared on, not by the content itself — “For the two datasets Science and Conspiracy and Vaccines, data were labeled in a binary way, respectively, provaccines/antivaccines and proscience/conspiracy, based on the page where they were posted.” | limitations[2] |
| limitation | interaction networks are reconstructed differently on each platform, and where direct relations are unavailable a proxy is assumed - a link between two users who commented on the same post — “In some cases, direct relations between users are not available in the data, so one needs to assume some proxy for social connections, for example, a link between two users if they comment on the same post on Facebook.” | limitations[3] |
| limitation | the differences across platforms could be an artefact of the different topics sampled, which is why one common topic had to be compared separately — “The striking differences observed across social media, in terms of homophily in the interaction networks and information diffusion, could be attributed to the different topics taken into account. For this reason, here we compare Facebook and Reddit on a common topic, news consumption.” | limitations[4] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | — “These results indicate that information diffusion is biased toward individuals who share a similar leaning in some social media, namely Twitter and Facebook. In contrast, in others—Reddit and Gab in our analysis—this effect is absent.” | content_tags.findings[0] |
| finding | — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | content_tags.findings[1] |
| finding | — “Such a latter configuration may depend upon two factors: 1) Gab and Reddit are not bursting the echo chamber effects, or 2) we are observing the dynamic inside a single echo chamber.” | content_tags.findings[2] |
| finding | — “One can see that all three measures confirm the picture obtained for other datasets: On Facebook, we observe a clear separation among users depending on their leaning, while, on Reddit, users’ leanings are more homogeneous and show only one peak.” | content_tags.findings[3] |
| finding | — “Note that polarization can happen independently from the structure or the very presence of social interactions.” | content_tags.findings[4] |
| quantity | media outlets carrying a political label, the base of every leaning score — “The labeling of news outlets is based on the information reported by Media Bias/Fact Check (MBFC) (https://mediabiasfactcheck.com), an independent fact-checking organization that rates news outlets on the basis of the reliability and of the political bias of the contents they produce and share. The labeling provided by MBFC, retrieved in June 2019, ranges from Extreme Left to Extreme Right for political bias. The total number of media outlets for which we have a political label is 2,190.” | content_tags.quantities[0] |
| quantity | unique pieces of content across the four platforms — “Overall, we account for the interactions of more than 1 million active users on the four platforms, for a total of more than 100 million unique pieces of content, including posts and social interactions.” | content_tags.quantities[1] |
| quantity | active users whose interactions are accounted for — “Overall, we account for the interactions of more than 1 million active users on the four platforms, for a total of more than 100 million unique pieces of content, including posts and social interactions.” | content_tags.quantities[2] |
| variable | homophily — “Then, we use these elements to assess echo chambers’ presence by looking at two different aspects: 1) homophily in interactions concerning a specific topic and 2) bias in information diffusion from like-minded sources.” | content_tags.variables[0] |
| variable | polarization — “Note that polarization can happen independently from the structure or the very presence of social interactions.” | content_tags.variables[1] |
| definition | SIR — “Classical epidemic models such as the susceptible–infected–recovered (SIR) model have been used to study the diffusion of information, such as rumors or news. In the SIR model, each agent can be in any of three states: susceptible (unaware of the circulating information), infectious (aware and willing to spread it further), or recovered (knowledgeable but not ready to transmit it anymore).” | content_tags.definitions[0] |
| definition | echo chambers — “We can broadly define echo chambers as environments in which the opinion, political leaning, or belief of users about a topic gets reinforced due to repeated interactions with peers or sources having similar tendencies and attitudes.” | content_tags.definitions[1] |
| definition | Selective exposure — “Selective exposure and confirmation bias (i.e., the tendency to seek information adhering to preexisting opinions) may explain the emergence of echo chambers on social media.” | content_tags.definitions[2] |
| definition | group polarization — “According to group polarization theory, an echo chamber can act as a mechanism to reinforce an existing opinion within a group and, as a result, move the entire group toward more extreme positions.” | content_tags.definitions[3] |
| concept | echo chambers — not defined in this paper's own words; environments where a user's opinion is reinforced by repeated interaction with similar peers - defined broadly here because the construct's very existence is contested — “We can broadly define echo chambers as environments in which the opinion, political leaning, or belief of users about a topic gets reinforced due to repeated interactions with peers or sources having similar tendencies and attitudes.” | concepts[0] |
| concept | observables — not defined in this paper's own words; what the two characterising elements are turned into so they can be quantified - the paper's own word for the move this corpus exists to record — “We introduce an operational definition of echo chambers to provide a common methodological ground to explore how different platforms influence their formation. In particular, we operationalize the two common elements that characterize echo chambers into observables that can be quantified and empirically measured, namely, 1) the inference of the user’s leaning for a specific topic (e.g., politics, vaccines) and 2) the structure of their social interactions on the platform.” | concepts[1] |
| concept | operational definition — not defined in this paper's own words; introduced explicitly to provide a common methodological ground across platforms, because without one the platforms cannot be compared at all — “We introduce an operational definition of echo chambers to provide a common methodological ground to explore how different platforms influence their formation. In particular, we operationalize the two common elements that characterize echo chambers into observables that can be quantified and empirically measured, namely, 1) the inference of the user’s leaning for a specific topic (e.g., politics, vaccines) and 2) the structure of their social interactions on the platform.” | concepts[2] |
| concept | homophily — not defined in this paper's own words; the correlation between a user's leaning and the average leaning of their neighbourhood - one of the two independent elements — “Then, we use these elements to assess echo chambers’ presence by looking at two different aspects: 1) homophily in interactions concerning a specific topic and 2) bias in information diffusion from like-minded sources.” | concepts[3] |
| concept | polarization — not defined in this paper's own words; concentration of the leaning distribution into clusters - and explicitly able to occur with no social interaction at all, which is why it must be measured separately — “Note that polarization can happen independently from the structure or the very presence of social interactions.” | concepts[4] |
| concept | group polarization — not defined in this paper's own words; the mechanism by which an echo chamber moves an entire group toward more extreme positions — “According to group polarization theory, an echo chamber can act as a mechanism to reinforce an existing opinion within a group and, as a result, move the entire group toward more extreme positions.” | concepts[5] |
| concept | Selective exposure — not defined in this paper's own words; with confirmation bias, the human tendency offered as the origin of echo chambers — “Selective exposure and confirmation bias (i.e., the tendency to seek information adhering to preexisting opinions) may explain the emergence of echo chambers on social media.” | concepts[6] |
| concept | Facebook — not defined in this paper's own words; no user-tweakable feed algorithm; shows the strongest segregation of the four platforms — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | concepts[7] |
| concept | Reddit — not defined in this paper's own words; feed algorithm tweakable by the user; shows a single community rather than opposed clusters — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | concepts[8] |
| concept | Gab — not defined in this paper's own words; low moderation, repeatedly suspended by its provider; like Reddit shows no dependence of influence-set leaning on user leaning, but biased to the right — “Gab claims to be a social platform aimed at protecting freedom of speech. However, low moderation and regulation on content has resulted in widespread hate speech. For these reasons, it has been repeatedly suspended by its service provider, and its mobile app has been banned from both App and Play stores.” | concepts[9] |
| concept | feed algorithm — not defined in this paper's own words; the discriminator the paper lands on: whether the user can tweak it separates the platforms that show echo chambers from those that do not — “Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithm tweakable by the users (e.g., Reddit) and social media that don’t provide such an option (e.g., Facebook and Twitter).” | concepts[10] |
| concept | Media Bias/Fact Check — not defined in this paper's own words; the external labelling of 2,190 outlets on which every user leaning in this study ultimately rests — “The labeling of news outlets is based on the information reported by Media Bias/Fact Check (MBFC) (https://mediabiasfactcheck.com), an independent fact-checking organization that rates news outlets on the basis of the reliability and of the political bias of the contents they produce and share. The labeling provided by MBFC, retrieved in June 2019, ranges from Extreme Left to Extreme Right for political bias. The total number of media outlets for which we have a political label is 2,190.” | concepts[11] |
| concept | interaction network — not defined in this paper's own words; the reconstructed graph over which information is assumed to flow - built differently on each platform, which is the comparison's weak joint — “Facebook and Reddit are particularly apt to a cross-comparison since they share the definition of individual leaning (computed by using the classification provided by mediabiasfactcheck.org; see Materials and Methods for further details) and the rationale in creating connections among users that is based on an interaction network.” | concepts[12] |
| SIR | — “Classical epidemic models such as the susceptible–infected–recovered (SIR) model have been used to study the diffusion of information, such as rumors or news. In the SIR model, each agent can be in any of three states: susceptible (unaware of the circulating information), infectious (aware and willing to spread it further), or recovered (knowledgeable but not ready to transmit it anymore).” | content_tags.statistics[0] |
Full text — tagged
The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.
# Abstract ### Significance We explore the key differences between the main social media platforms and how they are likely to influence information spreading and the formation of echo chambersterm. To assess the different dynamics, we perform a comparative analysis on more than 100 million pieces of content concerning controversial topics (e.g., gun control, vaccination, abortion) from Gabterm, Facebookterm, Redditterm, and Twitter. The analysis focuses on two main dimensions: 1) homophilyterm in the interaction networksterm and 2) bias in the information diffusion toward like-minded peers. Our results show that the aggregation in homophilic clusters of users dominates online dynamics. However, a direct comparison of news consumption on Facebookterm and Redditterm shows higher segregation on Facebookterm.
Social media may limit the exposure to diverse perspectives and favor the formation of groups of like-minded users framing and reinforcing a shared narrative, that is, echo chambersterm. However, the interaction paradigms among users and feed algorithmsterm greatly vary across social media platforms. This paper explores the key differences between the main social media platforms and how they are likely to influence information spreading and echo chambersterm’ formation. We perform a comparative analysis of more than 100 million pieces of content concerning several controversial topics (e.g., gun control, vaccination, abortion) from Gabterm, Facebookterm, Redditterm, and Twitter. We quantify echo chambersterm over social media by two main ingredients: 1) homophilyterm in the interaction networksterm and 2) bias in the information diffusion toward like-minded peers. Our results show that the aggregation of users in homophilic clusters dominate online interactions on Facebookterm and Twitter. We conclude the paper by directly comparing news consumption on Facebookterm and Redditterm, finding higher segregation on Facebookterm.
Social media radically changed the mechanism by which we access information and form our opinions. We need to understand how people seek or avoid information and how those decisions affect their behavior, especially when the news cycle—dominated by the disintermediated diffusion of information—alters the way information is consumed and reported on. A recent study limited to Twitter claimed that fake news travels faster than real news. However, a multitude of factors affects information spreading on social media platforms. Online polarizationterm, for instance, may foster misinformation spreading. Our attention span remains limited, and feed algorithmsterm might limit our selection process by suggesting contents similar to the ones we are usually exposed to. Furthermore, users show a tendency to favor information adhering to their beliefs and join groups formed around a shared narrative, that is, echo chambersterm. We can broadly define echo chambersterm as environments in which the opinion, political leaning, or belief of users about a topic gets reinforced due to repeated interactions with peers or sources having similar tendencies and attitudes. Selective exposureterm and confirmation bias (i.e., the tendency to seek information adhering to preexisting opinions) may explain the emergence of echo chambersterm on social media.
According to group polarizationterm theory, an echo chamberterm can act as a mechanism to reinforce an existing opinion within a group and, as a result, move the entire group toward more extreme positions. Echo chambersterm have been shown to exist in various forms of online media such as blogs, forums, and social media sites. Some studies point out echo chambersterm as an emerging effect of human tendencies, such as selective exposureterm, contagion, and group polarizationterm. However, recently, the effects and the very existence of echo chambersterm have been questioned. This issue is also fueled by the scarcity of comparative studies on social media, especially concerning news consumption. In this context, the debate around echo chambersterm is fundamental to understanding social media’s influence on information consumption and public opinion formation. In this paper, we explore the key differences between social media platforms and how they are likely to influence the formation of echo chambersterm or not. As recently shown in the case of selective exposureterm to news outlets, studies considering multiple platforms can offer a fresh view on long-debated problems. Different platforms offer different interaction paradigms to users, ranging from retweets and mentions on Twitter to likes and comments in groups on Facebookterm, thus triggering very different social dynamics. We introduce an operational definitionterm·mth of echo chambersterm to provide a common methodological ground to explore how different platforms influence their formation. In particular, we operationalize the two common elements that characterize echo chambersterm into observablesterm that can be quantified and empirically measured, namely, 1) the inference of the user’s leaning for a specific topic (e.g., politics, vaccines) and 2) the structure of their social interactions on the platform. Then, we use these elements to assess echo chambersterm’ presence by looking at two different aspects: 1) homophilyterm in interactions concerning a specific topic and 2) bias in information diffusion from like-minded sources. We focus our analysis on multiple platforms: Facebookterm, Twitter, Redditterm, and Gabterm. These platforms present similar features and functionalities (e.g., they all allow social feedback actions such as likes or upvotes) and design (e.g., Gabterm is similar to Twitter) but also distinctive features (e.g., Redditterm is structured in communities of interest called subreddits). Redditterm is one of the most visited websites worldwide (https://www.alexa.com/siteinfo/redditterm.com) and is organized as a forum to collect discussions on a wide range of topics, from politics to emotional support. Gabterm claims to be a social platform aimed at protecting freedom of speech. However, low moderation and regulation on content has resulted in widespread hate speech. For these reasons, it has been repeatedly suspended by its service provider, and its mobile app has been banned from both App and Play stores. Overall, we account for the interactions of more than 1 million active users on the four platforms, for a total of more than 100 million unique pieces of content, including posts and social interactions. Our analysis shows that platforms organized around social networks and news feed algorithmsterm, such as Facebookterm and Twitter, favor the emergence of echo chambersterm.
We conclude the paper by directly comparing news consumption on Facebookterm and Redditterm, finding higher segregation on Facebookterm than on Redditterm.
## Characterizing Echo Chambersterm in Social Media ### Operational Definitionsterm·mth.
To explore the key differences between social media platforms and how they influence echo chambersterm’ formation, we need to operationalize a definition for them. First, we need to identify the attitude of users at a microlevel. On online social media, the individual leaning of a user toward a specific topic, , can be inferred in different ways, via the content produced or the endorsement network among users. Concerning content, we can define the leaning as the attitude expressed by a piece of content toward a specific topic. This leaning can be explicit (e.g., arguments supporting a narrative) or implicit (e.g., framing and agenda setting). Let us consider a user producing a number of contents, , where is the activity of user , and each content leaning is assigned a numeric value. Then the individual leaning of user can be defined as the average of the leanings of produced contents,Once individual leanings have been inferred, polarizationterm can be defined as a state of the system such that the distribution of leanings, , is concentrated in one or more clusters. A possible example is the case of a single cluster, distinguishable by a single, extreme peak in . Another example is the typical case of topics characterized by positive versus negative stances, in which a bimodal distribution can describe polarizationterm. For instance, if opinions are assumed to be embedded in a one-dimensional space, without loss of generality, as usual for controversial topics, then polarizationterm is characterized by two well-separated peaks in , for positive and negative opinions. In contrast, neutral ones are absent or underrepresented in the population. Note that polarizationterm can happen independently from the structure or the very presence of social interactions. Homophilyterm in social interactions can be quantified by representing interactions as a social network and then analyzing its structure concerning the opinions of the users. Social networks can be reconstructed in different ways from online social media, where links represent social relationships or interactions. Since we are interested in capturing the possible exchange of opinions between users, we assume links as the substrate over which information may flow. For instance, if user follows user on Twitter, user can see tweets produced by user , and there is a flow of information from node to node in the network. When the reconstructed network is directed, we assume the link direction points to potential influencers (opposite of information flow). Actions such as mentions or retweets may convey similar flows. In some cases, direct relations between users are not available in the data, so one needs to assume some proxy for social connections, for example, a link between two users if they comment on the same post on Facebookterm. Crucially, the two elements characterizing the presence of echo chambersterm, polarizationterm and homophilic interactions, should be quantified independently.
### Implementation on Social Media.
This section explains how we implement the operational definitionsterm·mth defined above on different social media. For each medium, we detail 1) how we quantify users’ leaning, and 2) how we reconstruct how the information spread.
### Twitter.
We consider the set of tweets posted by user that contain links to news outlets of known political leaning. Each news outlet is associated with a political leaning score ranging from extreme left to extreme right following the Materials and Methods classification. We infer the individual leaning of a user, , , by averaging the news organizations’ scores linked by user according to Eq. 1. We analyze three different datasets collected on Twitter related to controversial topics: gun control, Obamacare, and abortion. For each dataset, the social interaction networkterm is reconstructed using the following relation so that there is a direct link from node to node if user follows user (i.e., the source). Henceforth, we focus on the dataset about abortion, and others are shown in SI Appendix.
### Facebookterm.
We quantify the individual leaning of users considering endorsements in the form of likes to posts. Posts are produced by pages that are labeled in a certain number of categories, and, to each category, we assign a numerical value (e.g., Anti-Vax [+1] or Pro-Vax [–1]). Each like to a post (only one like per post is allowed) represents an endorsement for that content, which is assumed to be aligned with the leaning associated with the page. Thus, the user’s leaning is defined as the average of the content leanings of the posts liked by the user, according to Eq. 1.
We analyze three different datasets collected on Facebookterm regarding a specific topic of discussion: vaccines, science versus conspiracy, and news. The interaction networkterm is defined by considering comments. In such an interaction networkterm, two users are connected if they cocommented on at least one post. Henceforth, we focus on the dataset about vaccines and news, and others are shown in SI Appendix.
### Redditterm.
The individual leaning of users is quantified similarly to Twitter by considering the links to news organizations in the content produced by the users, submissions, and comments. We build the interaction networkterm considering comments and submissions. There exists a direct link from node to node if user comments on a submission or comment by user (we assume that reads the comment they are replying to, which is written by ).
We analyze three datasets collected on different subreddits: the_donald, Politics, and News. In the following, we focus on the dataset collected on the Politics and the News subreddits, and others are shown in SI Appendix.
### Gabterm.
The political leaning of user is computed by considering the set of contents posted by user containing a link to news outlets of a known political leaning, similarly to Twitter and Redditterm. To obtain the leaning of user , we averaged the scores of each link posted by user according to Eq. 1. The interaction networkterm is reconstructed by exploiting the cocommenting relationships under posts in the same way as for Facebookterm. Given two users and , an undirected edge between and exists if and only if they comment under the same post.
## Comparative Analysis In the following, we perform a comparative analysis of four different social media. We select one dataset for each social media: Abortion (Twitter), Vaccines (Facebookterm), Politics (Redditterm), and Gabterm as a whole. Results for other datasets for the same medium are qualitatively similar, as shown in SI Appendix. We first characterize echo chambersterm in the networks’ topology, and then look at their effects on information diffusion. Finally, we directly compare news consumption on Facebookterm and Redditterm.
### Polarizationterm and Homophilyterm in the Interaction Networksterm.
The network’s topology can reveal echo chambersterm, where users are surrounded by peers with similar leanings, and thus they get exposed, with a higher probability, to similar contents. In network terms, this translates into a node with a given leaning more likely to be connected with nodes with a leaning close to . This concept can be quantified by defining, for each user , the average leaning of their neighborhood, as , where is the adjacency matrix of the interaction networkterm, if there is a link from node to node , otherwise, and is the out-degree of node . Fig. 1 shows the correlation between the leaning of a user and the leaning of their neighbors, , for the four social media under consideration. The probability distributions (individual leaning) and (average leaning of neighbors) are plotted on the and axes, respectively. All plots are color-coded contour maps, representing the number of users in the phase space : The brighter the area in the plan, the larger the density of users in that area. The topics of vaccines and abortion, on Facebookterm and Twitter, respectively, show a strong correlation between the leaning of a user and the average leaning of their nearest neighbors. Similar behavior is found for different topics from the same social media platform; see SI Appendix. Conversely, Redditterm and Gabterm show a different picture. The corresponding plots in Fig. 1 display a single bright area, indicating that users do not split into groups with opposite leaning but form a single community, biased to the left (Redditterm) or the right (Gabterm). Similar results are found for different datasets on Redditterm; see SI Appendix.
Joint distribution of the leaning of users and the average leaning of their neighborhood for different datasets. (A) Twitter, (B) Redditterm, (C) Facebookterm, and (D) Gabterm. Colors represent the density of users: The lighter the color, the larger the number of users. Marginal distribution and are plotted on the x and y axes, respectively. Facebookterm and Twitter present by homophilic clustering.
The presence of homophilic interactions can be confirmed by the community structure of the interaction networksterm. We detected communities by applying the Louvain algorithm, removing singleton communities with only one user. Then, we computed each community’s average leaning, determined as the average of individual leanings of its members. Fig. 2 shows the communities emerging for each social medium, arranged by increasing average leaning on the axis (color-coded from blue to red), while the axis reports the size of the community. On Facebookterm and Twitter, communities span the whole spectrum of possible leanings, but users with similar leanings form each community. Some communities are characterized by a robust average leaning, especially in the case of Facebookterm. These results are in accordance with the observation of homophilic interactions. Instead, communities on Redditterm and Gabterm do not cover the whole spectrum, and all show similar average leaning. Furthermore, the almost total absence of communities with leaning very close to 0 confirms the polarized state of the systems.
Size and average leaning of communities detected in different datasets. A and C show the full spectrum of leanings related to the topics of abortions and vaccines with regard to communities in B and D, where the political leaning is less sparse.
### Effects on Information Spreading.
Simple models of information spreading can gauge the presence of echo chambersterm: Users are expected to be more likely to exchange information with peers sharing a similar leaning. Classical epidemic models such as the susceptible–infected–recovered (SIRterm·def) model have been used to study the diffusion of information, such as rumors or news. In the SIRterm·def model, each agent can be in any of three states: susceptible (unaware of the circulating information), infectious (aware and willing to spread it further), or recovered (knowledgeable but not ready to transmit it anymore). Susceptible (unaware) users may become infectious (aware) upon contact with infected neighbors, with a specific transmission probability . Infectious users can spontaneously become recovered with probability . To measure the effects of the leaning of users on the diffusion of information, we run the SIRterm·def dynamics on the interaction networksterm, by starting the epidemic process with only one node infected, and stopping it when no more infectious nodes are left.
The set of nodes in a recovered state at the end of the dynamics started with user as a seed of infection, that is, those that become aware of the information initially propagated by user , form the set of influence of user , . Thus, the set of influence of a user represents those individuals that can be reached by a piece of content sent by him/her, depending on the effective infection ratio . One can compute the average leaning of the set of influence of user , , asThe quantity indicates how polarized the users are that can be reached by a message initially propagated by user .
Fig. 3 shows the average leaning of the influence sets reached by users with leaning , for the different datasets under consideration. The recovery rate is fixed at 0.2 for every dataset. In contrast, the ratio between the infection rate and average degree depends on the specific dataset and is reported in the caption of each figure.
Average leaning of the influence sets reached by users with leaning , for different datasets under consideration. Size and color of each point represent the average size of the influence sets. The parameters of the SIRterm·def dynamics are set to (A) , (B) , (C) , and (D) , while is fixed at 0.2 for all simulations.
Again, one can observe a clear distinction between Facebookterm and Twitter, on one side, and Redditterm and Gabterm on the other side. For the topics of vaccines and abortion, on Facebookterm and Twitter, respectively, users with a given leaning are much more likely to be reached by information propagated by users with similar leaning, that is, . Similar behavior is found for different topics from the same social media platform; see SI Appendix. Conversely, Redditterm and Gabterm show a different behavior: The average leaning of the set of influence, , does not depend on the leaning . As expected, the average leaning in these media is not zero. Still, it assumes negative (positive) values in Redditterm (Gabterm), indicating that the users of this platform are more likely to receive left (right)-leaning content.
These results indicate that information diffusion is biased toward individuals who share a similar leaning in some social media, namely Twitter and Facebookterm. In contrast, in others—Redditterm and Gabterm in our analysis—this effect is absent. Such a latter configuration may depend upon two factors: 1) Gabterm and Redditterm are not bursting the echo chamberterm effects, or 2) we are observing the dynamic inside a single echo chamberterm.
Our results are robust for different values of the effective infection ratio ; see SI Appendix. Furthermore, Fig. 3 shows that the spreading capacity, represented by the average size of the influence sets (color-coded in Fig. 3), depends on the leaning of the users. On Twitter, proabortion users are more likely to reach larger audiences. The same is true for antivax users on Facebookterm, left-leaning users on Redditterm, and right-leaning users on Gabterm (in this dataset, left-leaning users are almost absent).
### News Consumption on Facebookterm and Redditterm.
The striking differences observed across social media, in terms of homophilyterm in the interaction networksterm and information diffusion, could be attributed to the different topics taken into account. For this reason, here we compare Facebookterm and Redditterm on a common topic, news consumption. Facebookterm and Redditterm are particularly apt to a cross-comparison since they share the definition of individual leaning (computed by using the classification provided by mediabiasfactcheck.org; see Materials and Methods for further details) and the rationale in creating connections among users that is based on an interaction networkterm. Fig. 4 shows a direct comparison of news consumption on Facebookterm and Redditterm along the metrics used in the previous sections to quantify the presence of echo chambersterm: 1) the correlation between the leaning of a user and the average leaning of neighbors (Fig. 4, Top), 2) the average leaning of communities detected in the networks (Fig. 4, Middle), and 3) the average leaning of the influence sets reached by users with leaning , by running SIRterm·def dynamics (Fig. 4, Bottom). One can see that all three measures confirm the picture obtained for other datasets: On Facebookterm, we observe a clear separation among users depending on their leaning, while, on Redditterm, users’ leanings are more homogeneous and show only one peak. In the latter social media, even users displaying a more extreme leaning (noticeable in the marginal histogram of Fig. 4 B, Top) tend to interact with the majority. Moreover, on Facebookterm, the seed user’s leaning affects who the final recipients of the information are, therefore indicating the presence of echo chambersterm. On Redditterm, this effect is absent.
Direct comparison of news consumption on (A) Facebookterm and (B) Redditterm. Joint distribution of the leaning of users and the average leaning of their nearest neighbor (Top), size and average leaning of communities detected in the interaction networksterm (Middle), and average leaning of the influence sets reached by users with leaning , by running SIRterm·def dynamics (Bottom) with parameters for A, for B, and for both. Facebookterm presents a highly segregated structure with regard to Redditterm.
## Conclusions Social media platforms provide direct access to an unprecedented amount of content. Platforms originally designed for user entertainment changed the way information spread. Indeed, feed algorithmsterm mediate and influence the content promotion accounting for users’ preferences and attitudes. Such a paradigm shift affected the construction of social perceptions and the framing of narratives; it may influence policy making, political communication, and the evolution of public debate, especially on polarizing topics. Indeed, users online tend to prefer information adhering to their worldviews, ignore dissenting information, and form polarized groups around shared narratives. Furthermore, when polarizationterm is high, misinformation quickly proliferates.
Some argued that the veracity of the information might be used as a determinant for information spreading patterns. However, selective exposureterm dominates content consumption on social media, and different platforms may trigger very different dynamics. In this paper, we explore the key differences between the leading social media platforms and how they are likely to influence the formation of echo chambersterm and information spreading. To assess the different dynamics, we perform a comparative analysis on more than 100 million pieces of content concerning controversial topics (e.g., gun control, vaccination, abortion) from Gabterm, Facebookterm, Redditterm, and Twitter. The analysis focuses on two main dimensions: 1) homophilyterm in the interaction networksterm and 2) bias in the information diffusion toward like-minded peers. Our results show that the aggregation in homophilic clusters of users dominates online dynamics. However, a direct comparison of news consumption on Facebookterm and Redditterm shows higher segregation on Facebookterm. Furthermore, we find significant differences across platforms in terms of homophilic patterns in the network structure and biases in the information diffusion toward like-minded users. A clear-cut distinction emerges between social media having a feed algorithmterm tweakable by the users (e.g., Redditterm) and social media that don’t provide such an option (e.g., Facebookterm and Twitter). Our work provides important insights into the understanding of social dynamics and information consumption on social media. The next envisioned step addresses the temporal dimension of echo chambersterm, to understand better how different social feedback mechanisms, specific to distinct platforms, can impact their formation.
## Materials and Methods Here we provide details about the labeling of news outlets and the datasets considered.
### Labeling of Media Sources.
The labeling of news outlets is based on the information reported by Media Bias/Fact Checkterm (MBFC) (https://mediabiasfactcheck.com), an independent fact-checking organization that rates news outlets on the basis of the reliability and of the political bias of the contents they produce and share. The labeling provided by MBFC, retrieved in June 2019, ranges from Extreme Left to Extreme Right for political bias. The total number of media outlets for which we have a political label is 2,190. A detailed description of the source labeling process and political bias distribution can be found in SI Appendix.
### Data Availability.
For what concerns Gabterm, all data are available on the Pushshift public repository (https://pushshift.io/what-is-pushshift-io/) at this link: https://files.pushshift.io/gabterm/. Redditterm data are available on the Pushshift public repository at this link: https://search.pushshift.io/redditterm/. For what concerns Facebookterm and Twitter, we provide data according to their Terms of Services on the corresponding author institutional page at this link: https://walterquattrociocchi.site.uniroma1.it/ricerca. For news outlet classification, we used data from MBFC (https://mediabiasfactcheck.com), an independent fact-checking organization. Anonymized data have been deposited in Open Science Framework (10.17605/OSF.IO/X92BR). For further details about data, refer to the following section.
### Empirical Datasets.
Table 1 reports summary statistics of the datasets under consideration. Due to the structural differences among platforms, each dataset has different features. For Twitter, we used tweets regarding three topics collected by Garimella et al., namely Gun control, Obamacare, and abortion. Tweets linking to a news source with a known bias are classified based on MBFC. Facebookterm datasets were created by using Facebookterm Graph API and were previously explored in ref. (Science and Conspiracy), (Vaccines) and (News). For the two datasets Science and Conspiracy and Vaccines, data were labeled in a binary way, respectively, provaccines/antivaccines and proscience/conspiracy, based on the page where they were posted. Posts in the dataset News were instead classified based on MBFC labeling. Redditterm datasets have been obtained by downloading comments and submissions posted in the subreddit Politics, the_donald, and News and labeled according to the classification obtained from MBFC. The Gabterm dataset has been collected from https://files.pushshift.io/gabterm and contains posts, replies, and quotations. Posts were labeled according to MBFC classification. Further details can be found in SI Appendix.
Dataset details Media Dataset Twitter Gun control June 2016 14 d 19 million 3,963 0.93 Obamacare June 2016 7 d 39 million 8,703 0.90 Abortion June 2016 7 d 34 million 7,401 0.95 Facebookterm Sci/Cons January 2010 5 y 75,172 183,378 1.00 Vaccines January 2010 7 y 94,776 221,758 1.00 News January 2010 6 y 15,540 38,663 1.00 Redditterm Politics January 2017 1 y 353,864 240,455 0.15 the_donald January 2017 1 y 1.234 million 138,617 0.16 News January 2017 1 y 723,235 179,549 0.20 Gabterm Gabterm November 2017 1 y 13 million 165,162 0.13 For each dataset, we report the starting date of collection , time span expressed in days (d) or years (y), number of unique contents , number of users , and coverage (fraction of users with classified leaning). For Twitter, represents the window to sample active users, of which we retrieve all of the tweets related to the topic via the Application Programming Interface (API) (more information in SI Appendix). Sci/Cons, Scientific and Conspiracy content.
## Supplementary Material The authors declare no competing interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2023301118/-/DCSupplemental.