Political Polarization on Twitter
Open accessAbstract
In this study we investigate how social media shape the networked public sphere and facilitate communication between communities with different political orientations. We examine two networks of political communication on Twitter, comprised of more than 250,000 tweets from the six weeks leading up to the 2010 U.S. congressional midterm elections. Using a combination of network clustering algorithms and manually-annotated data we demonstrate that the network of political retweets exhibits a highly segregated partisan structure, with extremely limited connectivity between left- and right-leaning users. Surprisingly this is not the case for the user-to-user mention network, which is dominated by a single politically heterogeneous cluster of users in which ideologically-opposed individuals interact at a much higher rate compared to the network of retweets. To explain the distinct topologies of the retweet and mention networks we conjecture that politically motivated individuals provoke interaction by injecting partisan content into information streams whose primary audience consists of ideologically-opposed users. We conclude with statistical evidence in support of this hypothesis.
AI snapshot
Written from the paper’s full text.
TL;DR
Conover et al. analyze political communication on Twitter during the six weeks before the 2010 U.S. midterm elections, using retweet and mention networks built from 252,300 politically relevant tweets. They find that the retweet network is highly segregated along partisan lines, while the mention network is dominated by a single politically heterogeneous cluster, with cross-ideological interaction linked to hashtag content injection.
Key findings
- The retweet network is highly polarized: its modularity is 0.48 versus 0.17 for the mention network, and its modularity z-score against degree-preserving shuffled graphs is 11.02 versus 2.06 for mentions. In the retweet network, cluster A is 93.4% right-leaning and cluster B is 80.1% left-leaning.
- The mention network is politically heterogeneous: it is dominated by a single cluster accounting for more than 97% of users, and cross-ideological links are much more common than in retweets. Observed/expected link ratios for cross-ideological interaction are 0.68 (Left→Right) and 0.77 (Right→Left) in mentions, versus 0.05 and 0.03 in retweets.
- Manual annotation found most politically active users express a political identity: both annotators could not determine identity for only 8% of users, and more than 75% had unambiguous political valences when requiring both annotators to agree. Cohen’s Kappa was 0.80 for left and 0.82 for right, but only 0.42 for undecidable.
- Hashtag content injection helps explain the difference: users with neutral or mixed-valence hashtags receive and produce more inter-ideological mentions, and nearly 30% of tweets in the #p2 search feed originated from right-leaning users. Example valences include #tcot = 0.384 and #p2 = -0.605.
Methods
- Data were collected from Twitter’s gardenhose API between September 14 and November 1, 2010, observing approximately 355 million tweets. Political content was identified using seed hashtags #p2 and #tcot, a Jaccard similarity threshold of 0.005, 66 unique hashtags, and exclusion of 11 ambiguous hashtags, yielding 252,300 politically relevant tweets from more than 45,000 users.
- The authors constructed retweet and mention networks. The retweet network had 23,766 non-isolated nodes out of 45,365 total, with an 18,470-node largest connected component; the mention network had 10,142 non-isolated nodes out of 17,752 total, with a 7,175-node largest component. Analyses focused on the largest connected components.
- Community detection used label propagation seeded by leading-eigenvector modularity maximization for two communities, run 100 times and compared with Adjusted Rand Index. The authors compared observed modularity to 1,000 degree-preserving shuffled graphs, manually annotated 1,000 random users with a non-author judge on 200, and analyzed observed versus expected cross-ideological links and hashtag valence.
Limitations
- The paper states that direct modularity comparison is problematic because the retweet and mention networks differ in size and connectivity, so it uses shuffled graphs as a baseline; it also notes that the normality assumption for modularity values is false and uses Chebyshev’s inequality instead. It further states that cross-ideological interactions are “almost certainly not a panacea” and may reinforce polarization, and calls for further sentiment analysis and follower-network studies.
- Evident from the text: the sample is limited to six weeks before the 2010 U.S. midterm elections and to tweets containing a discovered set of political hashtags, so findings may not generalize to other periods, elections, or political contexts.
- Evident from the text: manual political alignment was based on 1,000 users, the undecidable category had only fair-to-moderate inter-annotator agreement (κ = 0.42), and the mention network is smaller than the retweet network, which may affect comparisons.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 10 cited sources checked · 0 supported · 0 flagged
- Findingown data
In the mention network, users placed in the same cluster are not likely to be much more similar to each other than users in different clusters.
“Figure 2 shows that in the mention network, users placed in the same cluster are not likely to be much more similar to each other than users in different clusters.”
Quote in the paperSubstantial claim - Findingown data
The community structure found in the retweet network is significantly more segregated than that found in the mention network.
“We conclude that the community structure found in the retweet network is significantly more segregated than that found in the mention network.”
Quote in the paperSubstantial claim - Findingown data
In both mention and retweet networks, one of the clusters is more cohesive than the other, meaning tag usage within one community is more homogeneous.
“Further, we note that in both mention and retweet networks, one of the clusters is more cohesive than the other -meaning the tag usage within one community is more homogeneous.”
Quote in the paperNot a substantial claim - Findingown data
All differences in average similarities between same-cluster and different-cluster user pairs in the retweet and mention networks are significant at the 95% confidence level.
“All differences are significant at the 95% confidence level.”
Quote in the paperSubstantial claim - Findingown data
A majority of politically active users on Twitter express a political identity in their tweets.
“Based on this analysis it is clear that a majority of politically active users on Twitter express a political identity in their tweets.”
Quote in the paperSubstantial claim
- Findingown data
The average similarity within retweet clusters is higher than across clusters.
“As a result the average similarity within retweet clusters is higher than across clusters.”
Quote in the paperSubstantial claim - Findingown data
The mention network is dominated by a politically heterogeneous cluster accounting for more than 97% of the users.
“Instead we find that it is dominated by a politically heterogeneous cluster accounting for more than 97% of the users”
Quote in the paperSubstantial claim - Findingown data
Messages in the style of the example tweets make up a substantial portion of the results for #tcot and #p2.
“messages in this style make up a substantial portion of the results.”
Quote in the paperSubstantial claim - Findingown data
Users contributing to a politically balanced combination of content streams on average receive and produce more inter-ideological communication than those who use mostly partisan hashtags.
“A key finding of this study, these results indicate that users contributing to a politically balanced combination of content streams on average receive and produce more inter-ideological communication than those who use mostly partisan hashtags.”
- Also stated in: § 5 Conclusions · hedged
“Specifically, users who apply hashtags with neutral or mixed valence are more likely to engage in communication with opposing communities.”
Quote in the paperSubstantial claim- Rests on an assumption: Any Twitter user can select arbitrary hashtags to annotate his or her tweets.
- Also stated in: § 5 Conclusions · hedged
- Findingown data
The content is frequently disparaging of the identities and views associated with users across the partisan divide.
“the content is frequently disparaging of the identities and views associated with users across the partisan divide.”
Quote in the paperSubstantial claim - Findingown data
Both annotators were unable to determine a political identity in only 8% of users.
“Both annotators were unable to determine a political identity in only 8% of users.”
Quote in the paperSubstantial claim - Findingown data
The retweet network is significantly more segregated than the mention network.
“We used these z-scores to argue that the retweet network is significantly more segregated than the mention network.”
- Also stated in: § Additional information · boosted
“we confirm that the retweet network is much more segregated.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The modularities of shuffled graph cluster assignments are normally distributed.
- Also stated in: § Additional information · boosted
- Findingown data
When a political identity is assigned only to users for whom both annotators agree, more than 75% of users have unambiguous political valences.
“if we assign a political identity only to users for whom both annotators agree, we report unambiguous political valences for more than 75% of users.”
Quote in the paperNot a substantial claim - Findingown data
Users frequently produce tweets containing hashtags that target multiple politically opposed audiences.
“We observe that users frequently produce tweets containing hashtags that target multiple politically opposed audiences”
Quote in the paperSubstantial claim - Findingown data
For both mentions and retweets, users are more likely to engage people with whom they agree.
“We see that for both means of communication, users are more likely to engage people with whom they agree.”
Quote in the paperSubstantial claim - Findingown data
The retweet network communities are maximally homogenous in the case of two clusters.
“Similarly, the retweet network communities are maximally homogenous in the case of two clusters.”
Quote in the paperSubstantial claim- Rests on an assumption: The partisan nature of U.S. political discourse makes two a natural number of clusters.
- Findingown data
The most popular hashtags do not have neutral valence.
“Moreover, Table 6 shows that the most popular hashtags do not have neutral valence”
Quote in the paperSubstantial claim - Findingown data
The network of political retweets exhibits a highly partisan community structure with two homogenous clusters of users who tend to share the same political identity.
“As shown in Table 4, the network of political retweets exhibits a highly partisan community structure with two homogenous clusters of users who tend to share the same political identity.”
Quote in the paperSubstantial claim- Rests on an assumption: The partisan nature of U.S. political discourse makes two a natural number of clusters.
- Findingown data
The mention network does not exhibit a clear partisan community structure.
“Surprisingly, the mention network does not exhibit a clear partisan community structure.”
- Also stated in: § 5 Conclusions · plain
“The retweet network is highly polarized, while the mention network is not.”
Quote in the paperSubstantial claim - Also stated in: § 5 Conclusions · plain
- Findingown data
The two major mechanisms for public political interaction on Twitter-mentions and retweets-induce distinct network topologies.
“In this study we have demonstrated that the two major mechanisms for public political interaction on Twitter -mentions and retweets -induce distinct network topologies.”
Quote in the paperSubstantial claim - Findingown data
The tendency to engage people with whom one agrees is far less pronounced in the mention network than in the retweet network.
“This effect, however, is far less pronounced in the mention network”
Quote in the paperSubstantial claim - Findingown data
In the mention network, significant amounts of cross-ideological interaction are observed.
“where we observe significant amounts of cross-ideological interaction.”
Quote in the paperSubstantial claim - Study design and conductown data
The simple left/right/undecidable coding definitions yielded the highest inter-annotator agreement in early trials compared with more detailed rubrics.
“We experimented with more detailed categorization rubrics but the simple definitions described above yielded the highest inter-annotator agreement in early trials of the coding process.”
Quote in the paperNot a substantial claim - Interpretationown data
The distinct topologies of the retweet and mention networks can be explained in part by politically motivated individuals injecting partisan content into information streams whose primary audience consists of ideologically opposed users.
“To explain the distinct topologies of the retweet and mention networks we conjecture that politically motivated individuals provoke interaction by injecting partisan content into information streams whose primary audience consists of ideologically-opposed users.”
- Also stated in: § 1 Introduction · hedged
“Finally, we provide evidence that these network structures result in part from politically motivated individuals annotating tweets with multiple hashtags whose primary audiences consist of ideologically-opposed users, a behavior also documented in the work of Yardi and boyd (2010).”
- Also stated in: § 1 Introduction · plain
“An interpretation of the observed community structures based on injection of partisan content into ideologically opposed hashtag information streams ( § 4).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Yardi and boyd (2010) source
- Rests on an assumption: Each hashtag identifies a stream of content, with users' tag choices denoting participation in different information channels.
- Also stated in: § 1 Introduction · hedged
- Interpretationown data
Mentions and replies may serve as a conduit through which users are exposed to information and opinions they might not choose in advance.
“This structural difference is of particular importance with respect to political communication, as we now have statistical evidence to suggest that mentions and replies may serve as a conduit through which users are exposed to information and opinions they might not choose in advance.”
Quote in the paperSubstantial claim- Rests on an assumption: When politically active individuals can avoid people and information they would not have chosen in advance, their opinions are likely to become increasingly extreme as a result of being exposed to more homogeneous viewpoints and fewer credible opposing opinions.
- Study design and conductown data
The cluster assignments produced by the label propagation method were highly consistent across 100 runs for each network, as measured by the Adjusted Rand Index.
“To confirm that we can produce consistent clusters across different runs we executed the algorithm one hundred times for each network and compared the label assignments produced by every run.”
- Also stated in: § 3.1 Community Structure · plain
“Table 3 reports the high average agreement between the resulting cluster assignments for each graph, as computed by the Adjusted Rand Index (Hubert and Arabie 1985).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Hubert and Arabie 1985 source
- Also stated in: § 3.1 Community Structure · plain
- Interpretationown data
This rules out that neutral-valence users are simply using the most popular hashtags.
“ruling out that neutral-valence users are simply using the most popular hashtags.”
Quote in the paperSubstantial claim - Interpretationreasoning
When engaging in political discourse, users often retweet just other users with whom they agree politically.
“The strong segregation evident in the retweet network and the fact that the two clusters correspond to political ideologies suggest that, when engaging in political discourse, users often retweet just other users with whom they agree politically.”
Quote in the paperSubstantial claim- Rests on an assumption: The partisan nature of U.S. political discourse makes two a natural number of clusters.
- Interpretationown data
The content of political discourse on Twitter remains highly partisan.
“Qualitatively speaking, our experience with this body of data suggests that the content of political discourse on Twitter remains highly partisan.”
Quote in the paperSubstantial claim - Interpretationreasoning
Individuals of different political alignments may interact with one another much more frequently using mentions.
“The dominance of the mention network by a single heterogeneous cluster of users, however, suggests that individuals of different political alignments may interact with one another much more frequently using mentions.”
Quote in the paperSubstantial claim - Interpretationreasoning
Political segregation, as manifested in the topology of the retweet network, persists in spite of substantial cross-ideological interaction.
“Whatever the case, political segregation, as manifested in the topology of the retweet network, persists in spite of substantial cross-ideological interaction.”
Quote in the paperSubstantial claim - Interpretationreasoning
The left-leaning hashtag #p21 appears to have emerged in response to efforts by right-leaning users to inject messages into the #p2 content stream.
“This tag appears to have emerged in response to the efforts by right-leaning users to inject messages into the high-profile #p2 content stream”
Quote in the paperSubstantial claim- Rests on an assumption: Any Twitter user can select arbitrary hashtags to annotate his or her tweets.
- Interpretationreasoning
This phenomenon of targeting multiple politically opposed audiences may be responsible in part for the network structures described in the study.
“we propose that this phenomenon may be responsible in part for the network structures described in this study.”
Quote in the paperSubstantial claim- Rests on an assumption: Any Twitter user can select arbitrary hashtags to annotate his or her tweets.
- Interpretationreasoning
Politically active Twitter users may be exposed to views with which they do not agree in the form of cross-ideological mentions.
“suggests that politically active Twitter users may be exposed to views with which they do not agree in the form of cross-ideological mentions.”
Quote in the paperSubstantial claim
- May combine two claims. 18
- Classified as probably not a substantial claim. 10
May combine two claims.
Findingown dataThe retweet and mention networks exhibit very similar scale-free topology (power-law degree distribution), with a number of users receiving or spreading a huge amount of information.
“We observe that the retweet and mention networks exhibit very similar scale-free topology (power-law degree distribution not shown), with a number of users receiving or spreading a huge amount of information.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataThe retweet network of political communication on Twitter in the six weeks before the 2010 U.S. midterm elections exhibited a highly modular, segregated partisan structure, with two homogenous communities corresponding to the political left and right and extremely limited connectivity between left- and right-leaning users.
“Using a combination of network clustering algorithms and manually-annotated data we demonstrate that the network of political retweets exhibits a highly segregated partisan structure, with extremely limited connectivity between left-and right-leaning users.”
- Also stated in: § 1 Introduction · boosted
“We demonstrate that the retweet network exhibits a highly modular structure, segregating users into two homogenous communities corresponding to the political left and right.”
- Also stated in: § 3.1 Community Structure · plain
“In summary, the retweet network contains two clusters of users who preferentially propagate content within their own communities.”
Quote in the paperSubstantial claim- Also stated in: § 1 Introduction · boosted
May combine two claims.
Findingown dataIn the retweet network, users in cluster A are more likely to have very similar profiles than users in cluster B, and users in different clusters are the least similar to each other.
“On the other hand, in the retweet network, users in cluster A are more likely to have very similar profiles than users in cluster B, and users in different clusters are the least similar to each other.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataThe retweet network exhibits two distinct communities of users, while the mention network is dominated by a single massive cluster of interconnected users.
“The retweet network exhibits two distinct communities of users, while the mention network is dominated by a single massive cluster of interconnected users.”
- Also stated in: § 3.1 Community Structure · plain
“Figure 1 shows the retweet and mention networks, laid out using a force-directed layout algorithm (Fruchterman and Reingold 1991), with node colors determined by the assigned communities.”
Quote in the paperSubstantial claim- Also stated in: § 3.1 Community Structure · plain
May combine two claims.
Findingown dataThe user-to-user mention network on Twitter in the same period did not exhibit political segregation; it was dominated by a single politically heterogeneous cluster in which ideologically opposed individuals interacted at a much higher rate than in the retweet network.
“Surprisingly this is not the case for the user-to-user mention network, which is dominated by a single politically heterogeneous cluster of users in which ideologically-opposed individuals interact at a much higher rate compared to the network of retweets.”
- Also stated in: § 1 Introduction · plain
“In contrast, we find that the mention network does not exhibit this kind of political segregation, resulting in users being exposed to individuals and information they would not have been likely to choose in advance.”
- Also stated in: § 1 Introduction · plain
“These data also show the mention network to be politically heterogeneous, with users of opposing political views interacting at a much higher rate than in the retweet network”
Quote in the paperSubstantial claim- Also stated in: § 1 Introduction · plain
May combine two claims.
Findingown dataThe retweet network clusters correspond to groups of users of similar political alignment, showing the retweet network is polarized on a partisan basis.
“Manual classification of Twitter users by political alignment, demonstrating that the retweet network clusters correspond to the political left and right.”
- Also stated in: § 3 Cluster Analysis · boosted
“by manually annotating users, we show that the retweet network is polarized on a partisan basis, while the mention network is much more politically heterogeneous.”
Quote in the paperSubstantial claim- Also stated in: § 3 Cluster Analysis · boosted
May combine two claims.
Findingown dataThe modularity induced by the communities in the retweet and mention networks was 0.48 and 0.17, respectively.
“The modularity induced by the communities in the retweet and mention networks have values of 0.48 and 0.17, respectively.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
FindingreasoningNearly 30% of the tweets in the #p2 search feed originate from right-leaning users.
“nearly 30% of the tweets in the #p2 search feed would originate from right-leaning users.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataIncreasing the number of target communities in the mention network does not reveal a more fine-grained ideological structure, but instead results in smaller yet politically heterogeneous clusters.
“Increasing the number of target communities in the mention network does not reveal a more fine-grained ideological structure, but instead results in smaller yet politically heterogeneous clusters.”
Quote in the paperSubstantial claimMay combine two claims.
Findingnone givenIdeologically-opposed users interact with one another, either through mentions or content injection, but they very rarely share information from across the divide with other members of their community.
“While we know for certain that ideologically-opposed users interact with one another, either through mentions or content injection, they very rarely share information from across the divide with other members of their community.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisereasoningRetweets and mentions serve distinct and complementary purposes, together acting as the primary mechanisms for explicit, public user-user interaction on Twitter.
“These two means of communication -retweets and mentions-serve distinct and complementary purposes, together acting as the primary mechanisms for explicit, public user-user interaction on Twitter.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationMentions function differently from retweets, allowing someone to address a specific user directly through the public feed, or, to a lesser extent, refer to an individual in the third person.
“Mentions function differently, allowing someone to address a specific user directly through the public feed, or, to a lesser extent, refer to an individual in the third person (Honeycutt and Herring 2008).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Honeycutt and Herring 2008
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Classified as probably not a substantial claim.
Study design and conductnone givenA non-author judge with broad knowledge of politics annotated 200 random users from the set of 1,000 to establish reproducibility of the annotation scheme.
“In line with the standards of the field, we had a non-author judge with a broad knowledge of politics annotate 200 random users from the set of 1,000 to establish the reproducibility of this annotation scheme.”
Quote in the paperSubstantial claimMay combine two claims.
Study design and conductcitationInter-annotator agreement between the 'left' and 'right' categories was κ = 0.80 and κ = 0.82, respectively, falling in the 'nearly perfect agreement' range.
“For agreement between the 'left' and 'right' categories we report κ = 0.80 and κ = 0.82 respectively, both of which fall in the "nearly perfect agreement" range (Landis and Koch 1977).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Landis and Koch 1977 source
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationRetweets act as a form of endorsement, allowing individuals to rebroadcast content generated by other users, thereby raising the content's visibility.
“Retweets act as a form of endorsement, allowing individuals to rebroadcast content generated by other users, thereby raising the content's visibility (boyd, Golder, and Lotan 2008).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites boyd, Golder, and Lotan 2008
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
May combine two claims.
Study design and conductown dataThe high agreement across runs suggests the clusters are consistent, so consensus clustering was avoided.
“Such a high agreement suggests that the clusters are consistent, and therefore we avoid resorting to consensus clustering for simplicity.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningThe injection of partisan content into ideologically opposed hashtag streams results in users being exposed to content they are not likely to rebroadcast, but to which they may respond using mentions.
“We argue that this process results in users being exposed to content they are not likely to rebroadcast, but to which they may respond using mentions, and provide statistical evidence in support of this hypothesis.”
Quote in the paperSubstantial claim- Rests on an assumption: Each hashtag identifies a stream of content, with users' tag choices denoting participation in different information channels.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationPolitical blogs preferentially link to other blogs of the same political ideology.
“Adamic and Glance (2005) famously demonstrated that political blogs preferentially link to other blogs of the same political ideology, a finding supported by the work of Hargittai, Gallo, and Kane (2007).”
Quote in the paperNot a substantial claim0 of 2 sources checkedMay combine two claims.
PremisecitationConsumers of online political information tend to choose to read blogs that share their political beliefs, with 26% more users doing so in 2008 than 2004.
“Consumers of online political information tend to behave similarly, choosing to read blogs that share their political beliefs, with 26% more users doing so in 2008than 2004(Pew Internet and American Life Project 2008).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Pew Internet and American Life Project 2008
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Classified as probably not a substantial claim.
PremisecitationPolitically active web users tend to organize into insular, homogenous communities segregated along partisan lines.
“Despite this, some empirical evidence suggests that politically active web users tend to organize into insular, homogenous communities segregated along partisan lines.”
Quote in the paperSubstantial claim0 of 2 sources checkedMay combine two claims.
Study design and conductown dataInter-annotator agreement for the undecidable category was fair to moderate (κ = 0.42), indicating that there are users for whom a political identity might be discernible in the context of specific domain knowledge.
“For the undecidable category we found "fair to moderate" agreement (κ = 0.42), indicating that there are users for whom a political identity might be discernible in the context of specific domain knowledge.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable).
Interpretationown dataMany messages contain sentiments more extreme than you would expect to encounter in face-to-face interactions.
“Many messages contain sentiments more extreme than you would expect to encounter in face-to-face interactions”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Interpretationnone givenWhen a user is exposed to ideologically opposed content, she will be unlikely to rebroadcast it, but may choose to respond directly to the originator in the form of a mention.
“We propose that when a user is exposed to ideologically opposed content in this way, she will be unlikely to rebroadcast it, but may choose to respond directly to the originator in the form of a mention.”
Quote in the paperNot a substantial claim- Rests on an assumption: Any Twitter user can select arbitrary hashtags to annotate his or her tweets.
Classified as probably not a substantial claim.
Interpretationreasoning#p21 ostensibly serves as a place where progressives can be exposed only to content aligned with their views.
“and ostensibly serves as a place where progressives can once again be exposed only to content aligned with their views.”
Quote in the paperSubstantial claim- Rests on an assumption: Any Twitter user can select arbitrary hashtags to annotate his or her tweets.
The paper relies on this without support.
When politically active individuals can avoid people and information they would not have chosen in advance, their opinions are likely to become increasingly extreme as a result of being exposed to more homogeneous viewpoints and fewer credible opposing opinions.
“The concern is that when politically active individuals can avoid people and information they would not have chosen in advance, their opinions are likely to become increasingly extreme as a result of being exposed to more homogeneous viewpoints and fewer credible opposing opinions.”
The paper relies on this without support.
A deliberative democracy relies on a broadly informed public and a healthy ecosystem of competing ideas.
“A deliberative democracy relies on a broadly informed public and a healthy ecosystem of competing ideas.”
The paper relies on this without support.
Each hashtag identifies a stream of content, with users' tag choices denoting participation in different information channels.
“Each hashtag identifies a stream of content, with users' tag choices denoting participation in different information channels.”
The authors state this as an assumption.
The modularities of shuffled graph cluster assignments are normally distributed.
“This discussion assumes that the modularities of the shuffled graph cluster assignments are distributed normally, which is not true in general.”
The paper relies on this without support.
The partisan nature of U.S. political discourse makes two a natural number of clusters.
“While the partisan nature of U.S. political discourse makes two a natural number of clusters”
The paper relies on this without support.
Any Twitter user can select arbitrary hashtags to annotate his or her tweets.
“Any Twitter user can select arbitrary hashtags to annotate his or her tweets.”
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Collection dates span about seven weeks, not six
MinorSeptember 14 to November 1 spans 48 days, or 49 days if both endpoints are included. Neither is six weeks.
“between September 14th and November 1st, 2010”
“During the six weeks of data collection”
Cites
The Measurement of Observer Agreement for Categorical Data
Landis, J. R., & Koch, G. G. (1977) · Biometrics · doi:10.2307/2529310 · Cited by 81,148 · cited 2× · see citing works
“For agreement between the 'left' and 'right' categories we report κ = 0.80 and κ = 0.82 respectively, both of which fall in the " nearly perfect agreement " range (Landis and Koch 1977).”
BackgroundInfluentialContent Analysis: An Introduction to its Methodology.
Shelley, M., & Krippendorff, K. (1984) · Journal of the American Statistical Association · doi:10.2307/2288384 · Cited by 24,305 · see citing works
Finding and evaluating community structure in networks
Newman, M. E. J., & Girvan, M. (2004) · Physical Review E · doi:10.1103/physreve.69.026113 · Cited by 13,931 · cited 2× · see citing works
“Modularity (Newman and Girvan 2004) resulting from the cluster assignments offers a first measure of segregation , and reinforces the qualitative finding above.”
ResultHubert, L., & Arabie, P. (1985) · Journal of Classification · doi:10.1007/bf01908075 · Cited by 8,219 · see citing works
Graph drawing by force‐directed placement
Fruchterman, T. M. J., & Reingold, E. M. (1991) · Software: Practice and Experience · doi:10.1002/spe.4380211102 · Cited by 6,352 · cited 1× · see citing works
“Figure 1 shows the retweet and mention networks, laid out using a force-directed layout algorithm (Fruchterman and Reingold 1991), with node colors determined by the assigned communities.”
MethodFinding community structure in networks using the eigenvectors of matrices
Newman, M. E. J. (2006) · Physical Review E · doi:10.1103/physreve.74.036104 · Cited by 4,799 · cited 2× · see citing works
“To improve its effectiveness and stability , we seeded the algorithm with initial node labels determined by the leading-eigenvector modularity maximization method for two clusters (Newman 2006).”
MethodNear linear time algorithm to detect community structures in large-scale networks
Raghavan, U. N., Albert, R., & Kumara, S. R. T. (2007) · Physical Review E · doi:10.1103/physreve.76.036106 · Cited by 3,535 · cited 1× · see citing works
“Label propagation (Raghavan, Albert, and Kumara 2007) works by assigning an initial arbitrary cluster membership to each node and then iteratively updating each node's label according to the label that is shared by most of its neighbors.”
BackgroundPredicting Elections with Twitter: What 140 Characters Reveal about Political Sentiment
Tumasjan, A., Sprenger, T. O., Sandner, P., & Welpe, I. M. (2010) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v4i1.14009 · Cited by 2,647 · cited 1× · see citing works
“Social media play an important role in shaping political discourse in the U.S. and around the world (Bennett 2003; Benkler 2006; Sunstein 2007; Farrell and Drezner 2008; Aday et al. 2010; Tumasjan et al. 2010; O'Connor et al. 2010).”
BackgroundThe political blogosphere and the 2004 U.S. election: divided they blog
Adamic, L. A., & Glance, N. (2005) · Proceedings of the 3rd international workshop on Link discovery · doi:10.1145/1134271.1134277 · Cited by 2,619 · see citing works
Tweet, Tweet, Retweet: Conversational Aspects of Retweeting on Twitter
Boyd, D., Golder, S., & Lotan, G. (2010) · 2010 43rd Hawaii International Conference on System Sciences · doi:10.1109/hicss.2010.412 · Cited by 2,051 · cited 1× · see citing works
“Retweets act as a form of endorsement, allowing individuals to rebroadcast content generated by other users, thereby raising the content's visibility (boyd, Golder, and Lotan 2008).”
BackgroundFrom Tweets to Polls: Linking Text Sentiment to Public Opinion Time Series
O'Connor, B., Balasubramanyan, R., Routledge, B., & Smith, N. A. (2010) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v4i1.14031 · Cited by 1,922 · cited 1× · see citing works
“and around the world (Bennett 2003; Benkler 2006; Sunstein 2007; Farrell and Drezner 2008; Aday et al. 2010; Tumasjan et al. 2010; O’Connor et al. 2010).”
BackgroundThe Wealth of Networks: How Social Production Transforms Markets and Freedom
(2006) · Choice Reviews Online · doi:10.5860/choice.44-0999 · Cited by 1,903 · see citing works
Kolbe, R. H., & Burnett, M. S. (1991) · Journal of Consumer Research · doi:10.1086/209256 · Cited by 1,516 · cited 4× · see citing works
“…annotations agree with an objective judge is Cohen's Kappa, defined as κ = P (α) − P () 1 − P () (2) where P (α) is the observed rate of agreement between annotators , and P () is the expected rate of random agreement given the relative frequency of each class label (Krippendorff 2004; Kolbe 1991).”
BackgroundMethodInfluentialThe law of group polarization.
Sunstein, C. R. (2002) · Journal of Political Philosophy · doi:10.1111/1467-9760.00148 · Cited by 1,371 · see citing works
Beyond Microblogging: Conversation and Collaboration via Twitter
Honeycutt, C., & Herring, S. (2009) · Hawaii International Conference on System Sciences · doi:10.1109/hicss.2009.602 · Cited by 1,055
Beyond Microblogging: Conversation and Collaboration via Twitter
Honey, C., & Herring, S. C. (2009) · Hawaii International Conference on System Sciences (HICSS) · doi:10.1109/hicss.2009.89 · Cited by 966 · see citing works
Dynamic debates: an analysis of group polarization over time on Twitter
Yardi, S., & Boyd, D. (2010) · Bulletin of Science, Technology & Society · doi:10.1177/0270467610380011 · Cited by 625 · cited 3× · see citing works
“Finally, we provide evidence that these network structures result in part from politically motivated individuals annotating tweets with multiple hashtags whose primary audiences consist of ideologically-opposed users, a behavior also documented in the work of Yardi and boyd (2010).”
BackgroundInfluentialAn Omnibus Test of Normality for Moderate and Large Size Samples
D’Agostino, R. B. (1971) · Biometrika · doi:10.1093/biomet/58.2.341 · Cited by 582 · cited 1× · see citing works
“By testing the sampled modularities with the omnibus K(2) statistic (D’Agostino 1971), we find that this assumption is not true: we have p < 10−10 that variations between the observed data and the best-fit normal distribution are due to random chance.”
MethodThe power and politics of blogs
Farrell, H. J., & Drezner, D. W. (2007) · Public Choice · doi:10.1007/s11127-007-9198-1 · Cited by 567 · cited 1× · see citing works
“Social media play an important role in shaping political discourse in the U.S. and around the world (Bennett 2003; Benkler 2006; Sunstein 2007; Farrell and Drezner 2008; Aday et al. 2010; Tumasjan et al. 2010; O'Connor et al. 2010).”
BackgroundRatkiewicz (2011) · doi:10.1145/1963192.1963301 · Cited by 541 · see citing works
Cross-ideological discussions among conservative and liberal bloggers
Hargittai, E., Gallo, J., & Kane, M. (2007) · Public Choice · doi:10.1007/s11127-007-9201-x · Cited by 246 · see citing works
Bennett, W. L. (2003) · Contesting Media Power · doi:10.5040/9798216409908.ch-2
Flagged first, then by citations
Research works printed but not linked: 1
- Sunstein, C. R. 2007. Republic.com 2.0. Princeton Univer- sity Press.
Cited by
Sorted by citations
Exposure to opposing views on social media can increase political polarization
Bail, C. A., Argyle, L. P., Brown, T. W., Bumpus, J. P., Chen, H., Hunzaker, M. B. F., Lee, J., Mann, M., Merhout, F., & Volfovsky, A. (2018) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1804840115 · Cited by 1,863 · see citing works
Emotions and Information Diffusion in Social Media—Sentiment of Microblogs and Sharing Behavior
Stieglitz, S., & Dang-Xuan, L. (2013) · Journal of Management Information Systems · doi:10.2753/mis0742-1222290408 · Cited by 1,604 · see citing works
“[18], who showed that users frequently produce tweets containing hashtags that table 4.”
BackgroundInfluentialTweeting From Left to Right: Is Online Political Communication More Than an Echo Chamber?
Barberá, P., Jost, J. T., Nagler, J., Tucker, J. A., & Bonneau, R. (2015) · Psychological Science · doi:10.1177/0956797615594620 · Cited by 1,472 · see citing works
“…but, until recently, estimates of individual-level characteristics of social-media users have been calculated only for small, self-selected samples (Conover et al., 2011; Kosinski, Stillwell, & Graepel, 2013) or have focused on variables that are of little direct relevance to the study of…”
BackgroundEmotion shapes the diffusion of moralized content in social networks
Brady, W. J., Wills, J. A., Jost, J. T., Tucker, J. A., & Van Bavel, J. J. (2017) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1618923114 · Cited by 1,396 · see citing works
The spread of low-credibility content by social bots
Shao, C., Ciampaglia, G. L., Varol, O., Yang, K.-C., Flammini, A., & Menczer, F. (2018) · Nature Communications · doi:10.1038/s41467-018-06930-7 · Cited by 1,125 · see citing works
“These include information overload and finite attention [41], novelty of false news [57], the selective exposure [52, 40, 39] caused by polarized and segregated online social networks [9, 8], algorithmic popularity bias [45, 20, 37], and other cognitive vulnerabilities such as confirmation bias and motivated reasoning…”
Self-citationBirds of the Same Feather Tweet Together: Bayesian Ideal Point Estimation Using Twitter Data
Barberá, P. (2015) · Political Analysis · doi:10.1093/pan/mpu011 · Cited by 982 · see citing works
“On one hand, Conover et al. (2010, 2011) find high levels of clustering along party lines: “the network of political retweets exhibits a highly segregated partisan structure, with extremely limited connectivity between left-and right-leaning users” (Conover et al. 2011, 89).”
BackgroundStatistical physics of vaccination
Wang, Z., Bauch, C. T., Bhattacharyya, S., d’Onofrio, A., Manfredi, P., Perc, M., Perra, N., Salathé, M., & Zhao, D. (2016) · Physics Reports · doi:10.1016/j.physrep.2016.10.006 · Cited by 954 · see citing works
Tucker, J. A., Guess, A., Barbera, P., Vaccari, C., Siegel, A., Sanovich, S., Stukal, D., & Nyhan, B. (2018) · doi:10.2139/ssrn.3144139 · Cited by 1,222 · see citing works
Social media and political communication: a social media analytics framework
Stieglitz, S., & Dang-Xuan, L. (2012) · Social Network Analysis and Mining · doi:10.1007/s13278-012-0079-3 · Cited by 649 · see citing works
“Conover et al. (2011) examine two networks of political communication on Twitter with more than 250,000 tweets from the 6 weeks leading up to the 2010 US congressional midterm elections.”
BackgroundThe spread of low-credibility content by social bots
Shao, C., Ciampaglia, G. L., Varol, O., Yang, K.-C., Flammini, A., & Menczer, F. (2018) · RePEc: Research Papers in Economics · Cited by 574 · see citing works
Self-citationInternet Research in Psychology
Gosling, S. D., & Mason, W. (2015) · Annual Review of Psychology · doi:10.1146/annurev-psych-010814-015321 · Cited by 458 · see citing works
Algorithmic Amplification of Politics on Twitter
Huszár, F., Ktena, S. I., O’Brien, C., Belli, L., Schlaikjer, A., & Hardt, M. (2022) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2025334119 · Cited by 65 · see citing works
Shao, C., Ciampaglia, G. L., Flammini, A., & Menczer, F. (2016) · International World Wide Web Conference (WWW) · doi:10.1145/2872518.2890098 · Cited by 410 · see citing works
“…features of the information exchange networks underlying social media create peculiar patterns of information access. homophily [25], polarization [12], algorithmic ranking [3], and social bubbles [28] — information environments with these challenges, here we present Hoaxy , an upcoming platform…”
BackgroundSelf-citationBadawy, A., Ferrara, E., & Lerman, K. (2018) · IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) · doi:10.1109/asonam.2018.8508646 · Cited by 353 · see citing works
“These effects were later documented by several studies [9, 15, 23, 25, 27, 46, 50, 56].”
BackgroundCommunity Interaction and Conflict on the Web
Kumar, S., Hamilton, W. L., Leskovec, J., & Jurafsky, D. (2018) · International World Wide Web Conference (WWW) · doi:10.1145/3178876.3186141 · Cited by 333 · see citing works
“forms has focused primarily on controversies [3, 13, 20, 21, 45, 52] and anti-social behavior, in the form of trolling [11, 12, 42], sockpuppetry [39], harassment and cyberbullying [8, 23, 32, 35, 55, 66, 79], vandalism [43], hate speech [9, 15, 51, 56, 72], and others [17, 24, 40, 41, 44, 46, 54, 68, 78, 81].”
BackgroundGenerative Adversarial Networks for face generation: A survey
Kammoun, A., Slama, R., Tabia, H., Ouni, T., & Abid, M. (2022) · ACM Computing Surveys · doi:10.1145/1122445.1122456 · Cited by 326 · see citing works
Political Astroturfing on Twitter: How to Coordinate a Disinformation Campaign
Keller, F. B., Schoch, D., Stier, S., & Yang, J. (2020) · Political Communication · doi:10.1080/10584609.2019.1661888 · Cited by 378 · see citing works
“Retweeting is often used to amplify the impact of a campaign (Yang & Kim, 2017), and retweet patterns usually reflect political divisions among Twitter users (Conover et al., 2011).”
BackgroundQuantitative Approaches to Comparing Communication Patterns on Twitter
Bruns, A., & Stieglitz, S. (2012) · Journal of Technology in Human Services · doi:10.1080/15228835.2012.744249 · Cited by 249 · see citing works
“Conover et al. (2011) examine two networks of political communication on Twitter with more than 250,000 tweets from the 6 weeks leading up to the 2010 U.”
BackgroundMeasuring political polarization: Twitter shows the two sides of Venezuela
Morales, A. J., Borondo, J., Losada, J. C., & Benito, R. M. (2015) · Chaos An Interdisciplinary Journal of Nonlinear Science · doi:10.1063/1.4913758 · Cited by 219 · see citing works
“In fact political segregation has already been observed on political blogs [29] or Twitter [30, 31].”
BackgroundClassifying Political Orientation on Twitter: It’s Not Easy!
Cohen, R., & Ruths, D. A. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v7i1.14434 · Cited by 208 · see citing works
“While this method did not achieve the best performance to date (93% vs. 95%), it was benchmarked on a less restrictive dataset than (Conover et al. 2011b) which likely contributed to the small different in performance.”
BackgroundMethodInfluentialPolitical Effects of the Internet and Social MediaPreprint
Zhuravskaya, E., Petrova, M., & Enikolopov, R. (2020) · Annual Review of Economics · doi:10.2139/ssrn.3439957 · Cited by 480 · see citing works
“Similarly, Conover et al. (2011) study links between political tweets in the six weeks leading up to the 2010 U.”
A Measure of Polarization on Social Media Networks Based on Community Boundaries
Guerra, P., Meira, W., Cardie, C., & Kleinberg, R. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v7i1.14421 · Cited by 188 · see citing works
“01 higher than the value that (Conover et al. 2011) found to conclude that Q = 0.”
BackgroundResultInfluentialInformation gerrymandering and undemocratic decisions
Stewart, A. J., Mosleh, M., Diakonova, M., Arechar, A. A., Rand, D. G., & Plotkin, J. B. (2019) · Nature · doi:10.1038/s41586-019-1507-6 · Cited by 176 · see citing works
“j=i N/2+Zp ✓ N/2 Zp i j ◆ q j p ✓ N/2 Zy j ◆ (1 qy) j (7)”
Limited individual attention and online virality of low-quality information
Qiu, X., Oliveira, D. F. M., Shirazi, A. S., Flammini, A., & Menczer, F. (2017) · Nature Human Behaviour · doi:10.1038/s41562-017-0132 · Cited by 186 · see citing works
“This may further lead to polarization [57, 13, 56, 43]; one group may automatically discount ideas from another [39, 44].”
BackgroundSelf-citationRetractedA Long-Term Analysis of Polarization on Twitter
Garimella, V. R. K., & Weber, I. G. (2017) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v11i1.14918 · Cited by 189 · see citing works
The Geospatial Characteristics of a Social Movement Communication Network
Conover, M., Davis, C. A., Ferrara, E., McKelvey, K., Menczer, F., & Flammini, A. (2013) · PLoS ONE · doi:10.1371/journal.pone.0055957 · Cited by 152 · see citing works
“Among others, it has been used to study influence and credibility [22–26], social structure [27–29] and to monitor users’ sentiment [30–33].”
BackgroundSelf-citationFalling into the Echo Chamber: The Italian Vaccination Debate on Twitter
Cossard, A., De Francisci Morales, G., Kalimeri, K., Mejova, Y., Paolotti, D., & Starnini, M. (2020) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v14i1.7285 · Cited by 126 · see citing works
“Indeed, Conover et al. (2011) show that mentions are often used to speak to (and attack) the opposing side.”
BackgroundInfluentialSocial Media Polarization and Echo Chambers in the Context of COVID-19: Case Study
Jiang, J., Ren, X., & Ferrara, E. (2021) · JMIRx Med · doi:10.2196/29570 · Cited by 135 · see citing works
“Social media platforms have been criticized for enhancing political echo chambers and driving political polarization [8-10].”
BackgroundMethodInfluentialMe, My Echo Chamber, and I: Introspection on Social Media Polarization
Gillani, N., Yuan, A., Saveski, M., Vosoughi, S., & Roy, D. (2018) · Proceedings of the 2018 World Wide Web Conference on World Wide Web - WWW '18 · doi:10.1145/3178876.3186130 · Cited by 117 · see citing works
“ed mention networks — suggesting homophilous endorsement patterns accompanied by a tendency for politically active users to inject partisan content and “call out” their counterparts on the other side [7]. Therefore, even when social media platforms create opportunities for cross-cutting dialog, the dialog is often muc…”
BackgroundInfluentialDoes Social Media Influence Conflict? Evidence from the 2012 Gaza Conflict
Zeitzoff, T. (2016) · Journal of Conflict Resolution · doi:10.1177/0022002716650925 · Cited by 115 · see citing works
Twitter in Politics: A Comprehensive Literature ReviewPreprint
Jungherr, A. M. (2014) · SSRN Electronic Journal · doi:10.2139/ssrn.2402443 · Cited by 112 · see citing works
Weakly Supervised User Profile Extraction from Twitter
Li, J., Ritter, A., & Hovy, E. H. (2014) · Annual Meeting of the Association for Computational Linguistics (ACL) · doi:10.3115/v1/p14-1016 · Cited by 112 · see citing works
“, 2010), or political polarity (Pennacchiotti and Popescu, 2011; Conover et al., 2011).”
BackgroundCharacterizing interactions in online social networks during exceptional events
Omodei, E., Domenico, M. D., & Arenas, À. (2015) · Frontiers in Physics · doi:10.3389/fphy.2015.00059 · Cited by 114 · see citing works
“sing three dierent actions, namely mentioning, replying and retweeting. While some evidences have shown that users tend to exploit in dierent ways the actions made available by the Twitter platform [20], such dierences have not been quantied so far. In this work, we analyze the activities of users from a new perspecti…”
BackgroundDo social media undermine social cohesion? A critical review
González-Bailón, S., & Lelkes, Y. (2022) · Social Issues and Policy Review · doi:10.1111/sipr.12091 · Cited by 110 · see citing works
Quantifying Political Leaning from Tweets, Retweets, and Retweeters
Wong, F. M. F., Tan, C. W., Sen, S., & Chiang, M. (2016) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2016.2553667 · Cited by 103 · see citing works
“Partisanship also increases with vocalness of the population.”
Background#MaskOn! #MaskOff! Digital polarization of mask-wearing in the United States during COVID-19
Lang, J., Erickson, W. W., & Jing-Schmidt, Z. (2021) · PLoS ONE · doi:10.1371/journal.pone.0250817 · Cited by 104 · see citing works
Social Media Networks, Fake News, and Polarization
Azzimonti, M., & Fernandes, M. (2018) · National Bureau of Economic Research · doi:10.3386/w24462 · Cited by 99 · see citing works
“Examples are DeGroot (1974), Ellison and Fundenberg (1993, 1995), Bala and Goyal (1998,2001), De Marzo, Vayanos and Zwiebel (2003) and Golub and Jackson (2010).”
BackgroundModeling the emergence of affective polarization in the social media society
Törnberg, P., Andersson, C., Lindgren, K., & Banisch, S. (2021) · PLOS ONE · doi:10.1371/journal.pone.0258259 · Cited by 118 · see citing works
Ruffo, G., Semeraro, A., Giachanou, A., & Rosso, P. (2022) · Computer Science Review · doi:10.1016/j.cosrev.2022.100531 · Cited by 110 · see citing works
“[55] looked at 250,000 tweets before the 2010 U.”
BackgroundInfluentialPeople Are Strange When You're a Stranger: Impact and Influence of Bots on Social Networks
Aiello, L. M., Deplano, M., Schifanella, R., & Ruffo, G. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v6i1.14236 · Cited by 97 · see citing works
“An additional evidence of user polarization into two opinion clusters can be given by applying community detection algorithms to the networks and check whether the detected clusters match the group membership of nodes (Conover et al. 2011).”
BackgroundHannak, A., Margolin, D., Keegan, B., & Weber, I. (2014) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v8i1.14555 · Cited by 90 · see citing works
“This follower network exhibits a classic polarization (Adamic and Glance 2005; Conover et al. 2011) between two denselyknit communities.”
BackgroundIdeological and Temporal Components of Network Polarization in Online Political Participatory Media
García, D., Abisheva, A., Schweighofer, S., Serdült, U., & Schweitzer, F. (2015) · Policy & Internet · doi:10.1002/poi3.82 · Cited by 91 · see citing works
“Future works can include datasets from the electorate at large, measuring mass polarization from other digital traces such as blogs (Adamic and Glance, 2005) and Twitter (Conover et al., 2011).”
BackgroundInfluentialThe Twitter parliamentarian database: Analyzing Twitter politics across 26 countries
van Vliet, L., Törnberg, P., & Uitermark, J. (2020) · PLoS ONE · doi:10.1371/journal.pone.0237073 · Cited by 93 · see citing works
Predicting the Topical Stance and Political Leaning of Media using Tweets
Stefanov, P., Darwish, K. M., Atanasov, A., & Nakov, P. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.50 · Cited by 82 · see citing works
“The original valence score described by Conover et al. (2011a) is calculated as follows: where t f ( u , C 0 ) is the number of times (term frequency) item u is cited by group C 0 , and total ( C 0 ) is the sum of the term frequencies of all items cited by C 0 . t f ( u , C 1 ) and total ( C 1 )…”
BackgroundMethodQuantifying Political Polarity Based on Bipartite Opinion Networks
Akoglu, L. (2014) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v8i1.14524 · Cited by 79 · see citing works
“2011) and (Conover et al. 2011b) both use modularity (Newman 2006) as a measure of segregation between political groups in Twitter.”
BackgroundInfluentialQuantifying Political Leaning from Tweets and Retweets
Wong, F. M. F., Tan, C. W., Sen, S., & Chiang, M. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v7i1.14422 · Cited by 76 · see citing works
“Regarding Twitter, political polarization was studied in (Conover et al. 2011b).”
BackgroundMethodSemantic Sentiment Analysis in Arabic Social Media
Tartir, S., & Abdul-Nabi, I. (2016) · Journal of King Saud University - Computer and Information Sciences · doi:10.1016/j.jksuci.2016.11.011 · Cited by 75 · see citing works
“The research in (Conover et al., 2011) showed that Twitter can be used as a platform for political deliberation.”
Quantifying ideological polarization on a network using generalized Euclidean distance
Hohmann, M., Devriendt, K., & Coscia, M. (2023) · Science Advances · doi:10.1126/sciadv.abq2044 · Cited by 74 · see citing works
Community-Based Fact-Checking on Twitter’s Birdwatch Platform
Pröllochs, N. (2022) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v16i1.19335 · Cited by 68 · see citing works
Mahl, D., Zeng, J., & Schäfer, M. S. (2021) · Social Media + Society · doi:10.1177/20563051211017482 · Cited by 67 · see citing works
“As prior literature indicates (e.g., Conover et al., 2011; Graham, 2016), hashtags affiliated with conservative politics (e.g., #TheGreatAwakening, #WWG1WGA, #tcot, #tgdn) are commonly used by their supporters for self-identification, regardless of the actual topic of the tweet.”
BackgroundGallacher, J., Heerdink, M. W., & Hewstone, M. (2021) · Social Media + Society · doi:10.1177/2056305120984445 · Cited by 66 · see citing works
“0% of interactions occur between ingroup members (Conover et al., 2011).”
BackgroundResultInfluentialEvent Detection via Communication Pattern Analysis
Chierichetti, F., Kleinberg, J. M., Kumar, R., Mahdian, M., & Pandey, S. (2014) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v8i1.14536 · Cited by 66 · see citing works
“Yardi and boyd (2010) and Conover et al. (2011) studied the effect of polarization on Twitter. For some early papers investigating the social network of Twitter, see the work of Java et al. (2007) and Krishnamurthy, Gill, and Arlitt (2008).”
BackgroundNeutral bots probe political bias on social media
Chen, W., Pacheco, D. F., Yang, K.-C., & Menczer, F. (2021) · RePEc: Research Papers in Economics · Cited by 68 · see citing works
Self-citationNeutral bots probe political bias on social media
Chen, W., Pacheco, D. F., Yang, K.-C., & Menczer, F. (2021) · Nature Communications · doi:10.1038/s41467-021-25738-6 · Cited by 66 · see citing works
Self-citationDoes populism go viral? How Italian leaders engage citizens through social media
Bracciale, R., Andretta, M., & Martella, A. (2021) · Information Communication & Society · doi:10.1080/1369118x.2021.1874472 · Cited by 64 · see citing works
“Finally, the strong polarization, which is often amplified on social media (Conover et al., 2011; Guimaraes et al., 2017), combines with an adversarial use of online platforms to create another typical trait of populist communication (van Kessel & Castelein, 2016; Waisbord & Amado, 2017).”
BackgroundCohesion and Coalition Formation in the European Parliament: Roll-Call Votes and Twitter Activities
Cherepnalkoski, D., Karpf, A., Mozetič, I., & Grćar, M. (2016) · PLoS ONE · doi:10.1371/journal.pone.0166586 · Cited by 64 · see citing works
Xie, P., Chen, H., & Hu, Y. J. (2020) · Journal of Management Information Systems · doi:10.1080/07421222.2020.1831762 · Cited by 62 · see citing works
Social Politics: Agenda Setting and Political Communication on Social Media
Yang, X., Chen, B.-C., Maity, M., & Ferrara, E. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-47880-7_20 · Cited by 65 · see citing works
“[12] found that the network of political retweets exhibits a highly segregated bipartisan structure, which seems to reflect the users’ political leanings, similarly to political blogs [1].”
Social media hypes about agro-food issues: Activism, scandals and conflicts
Stevens, T. M., Aarts, N., Termeer, C. J. A. M., & Dewulf, A. R. P. J. (2018) · Food Policy · doi:10.1016/j.foodpol.2018.04.009 · Cited by 63 · see citing works
Examining Sentiments and Popularity of Pro- and Anti-Vaccination Videos on YouTube
Song, M. Y., & Gruzd, A. A. (2017) · International Conference on Social Media and Society (SMSociety) · doi:10.1145/3097286.3097303 · Cited by 62 · see citing works
“It would be useful to apply concepts in political research such as the polarization of public opinion [86]–[88] to look for new ways to approach and understand the dichotomized vaccine sentiment.”
BackgroundAlmaatouq, A., Shmueli, E., Nouh, M., Alabdulkareem, A., Singh, V. K., Alsaleh, M., Alarifi, A., Alfaris, A., & Pentland, A. (2016) · International Journal of Information Security · doi:10.1007/s10207-016-0321-5 · Cited by 65 · see citing works
“With the rapid growth of OSNs popularity, we are witnessing an increased usage of these services to discuss issues of public interest and hence shape public opinions [16].”
BackgroundProceedings of the First Workshop on Abusive Language Online
(2017) · doi:10.18653/v1/w17-30 · Cited by 63 · see citing works
A model of opinion and propagation structure polarization in social media
Prasetya, H. A., & Murata, T. (2020) · Computational Social Networks · doi:10.1186/s40649-019-0076-z · Cited by 61 · see citing works
“In recent years, multiple empirical studies have revealed the existence of segregated communities in online social media interaction [4, 7, 9, 14, 18, 39, 73], where likeminded people interact with little exposure to different viewpoints.”
BackgroundResultEnvironmental Decision-Making in Times of Polarization
Judge, M., Kashima, Y., Steg, L., & Dietz, T. M. (2023) · Annual Review of Environment and Resources · doi:10.1146/annurev-environ-112321-115339 · Cited by 72 · see citing works
Li, J., & Su, M.-H. (2020) · Social Media + Society · doi:10.1177/2056305120916841 · Cited by 60 · see citing works
Asymmetrical perceptions of partisan political bots
Yan, H. Y., Yang, K.-C., Menczer, F., & Shanahan, J. E. (2020) · New Media & Society · doi:10.1177/1461444820942744 · Cited by 60 · see citing works
“The results have important implications for connecting the two theories (Lelkes and Westwood, 2017) and explaining political polarization on Twitter (Conover et al., 2011).”
BackgroundSelf-citationElecciones Europeas 2014: Viralidad de los mensajes en Twitter
Congosto, M. (2015) · Redes Revista hispana para el análisis de redes sociales · doi:10.5565/rev/redes.529 · Cited by 59 · see citing works
#bias: Measuring the Tweeting Behavior of Propagandists
Lumezanu, C., Feamster, N., & Klein, H. K. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v6i1.14247 · Cited by 59 · see citing works
“To classify a user as biased or neutral, we apply the following finding of Conover et al. (Conover et al. 2010), used to find political polarization on Twitter: users with similar ideologies tend to retweet exclusively each other's messages .”
BackgroundMethodInfluentialRizoiu, M.-A., Graham, T., Zhang, R., Zhang, Y., Ackland, R., & Xie, L. (2018) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v12i1.15029 · Cited by 60 · see citing works
“4.1), and the extreme bi-modality can be explained by the clear partisan nature of the chosen hashtags and by the known political polarization of users on Twitter (Conover et al. 2011; Barberá et al. 2015), which will be greatly enhanced in the context of a political debate.”
BackgroundInfluentialMeasuring Social Spam and the Effect of Bots on Information Diffusion in Social Media
Ferrara, E. (2018) · Computational social sciences · doi:10.1007/978-3-319-77332-2_13 · Cited by 57 · see citing works
“This is a very strict and conservative partisanship assignment, likely less prone to misclassification that may be yield by automatic machine-learning techniques not based on manual validation, e.g., [21].”
Sentiment Analysis of Arabic Slang Comments on Facebook
Soliman, T. H. A., Elmasry, M. A., Hedar, A.-R., et al. (2014) · INTERNATIONAL JOURNAL OF COMPUTERS & TECHNOLOGY · doi:10.24297/ijct.v12i5.2917 · Cited by 54 · see citing works
“Conover et al. [22] examined the retweet network of 250,000 political tweets during the six weeks prior to the 2010 U.S. midterm elections.”
BackgroundFrom Brexit to Trump: Social Media’s Role in Democracy
Hall, W., Tinati, R., & Jennings, W. (2018) · Computer · doi:10.1109/mc.2018.1151005 · Cited by 53 · see citing works
Stark, B., Magin, M., & Jürgens, P. (2021) · Springer eBooks · doi:10.1007/978-3-658-32133-8_17 · Cited by 57 · see citing works
Consensus and polarization in competing complex contagion processes
de Vasconcelos, V. V. L., Levin, S. A., & Pinheiro, F. L. (2019) · Journal of The Royal Society Interface · doi:10.1098/rsif.2019.0196 · Cited by 50 · see citing works
“[9] Michael Conover, Jacob Ratkiewicz, Matthew R Francisco, Bruno Gonçalves, Filippo Menczer, and Alessandro Flammini.”
BackgroundUsing Twitter to track immigration sentiment during early stages of the COVID-19 pandemic
Rowe, F., Mahony, M., Graells-Garrido, E., Rango, M., & Sievers, N. (2021) · Data & Policy · doi:10.1017/dap.2021.38 · Cited by 50 · see citing works
“Evidence of increasing political polarisation exists in the United Kingdom and the United States, and immigration has featured as a main divisive topic (Conover et al., 2011; Hong and Kim, 2016; Bail et al., 2018).”
BackgroundAccuracy and Political Bias of News Source Credibility Ratings by Large Language Models
Yang, K.-C., & Menczer, F. (2025) · Proceedings of the 17th ACM Web Science Conference 2025 · doi:10.1145/3717867.3717903 · Cited by 81 · see citing works
“When handling political information, the inherent partisan biases in LLMs could exacerbate echo chambers and polarization (Cinelli et al. 2021; Conover et al. 2021).”
Self-citationEmbeddings-Based Clustering for Target Specific Stances: The Case of a Polarized Turkey
Rashed, A. A.-A., Kutlu, M., Darwish, K. M., Elsayed, T., & Bayrak, C. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v15i1.18082 · Cited by 49 · see citing works
“We assign prominence scores for terms using valence scores (Conover et al. 2011) and term frequencies.”
MethodInfluentialSustained online amplification of COVID-19 elites in the United States
Gallagher, R. J., Doroshenko, L., Shugars, S., Lazer, D., & Foucault Welles, B. (2021) · Social Media + Society · doi:10.1177/20563051211024957 · Cited by 80 · see citing works
“This can lead to a balkanization of information, where different subgroups collectively amplify very different information sources and crowdsource very different elites (Conover et al. 2011; Foucault Welles and Jackson 2019).”
BackgroundTwitter-Based Analysis of the Dynamics of Collective Attention to Political Parties
Eom, Y.-H., Puliga, M., Smailović, J., Mozetič, I., & Caldarelli, G. (2015) · PLoS ONE · doi:10.1371/journal.pone.0131184 · Cited by 48 · see citing works
Are Echo Chambers Based on Partisanship? Twitter and Political Polarity in Poland and Hungary
Matuszewski, P., & Szabó, G. (2019) · Social Media + Society · doi:10.1177/2056305119837671 · Cited by 37 · see citing works
“One strand of the literature explores information-seeking patterns by looking at the correlation between exposure and the topological or geographical properties of messages (boyd, Golder, & Lotan, 2010; Conover et al., 2011; Java, Song, Finin, & Tseng, 2007).”
BackgroundOnline clustering, fear and uncertainty in Egypt’s transition
Lynch, M., Freelon, D., & Aday, S. (2017) · Democratization · doi:10.1080/13510347.2017.1289179 · Cited by 46 · see citing works
Tubes and bubbles topological confinement of YouTube recommendations
Roth, C., Mazières, A., & Menezes, T. (2020) · PLoS ONE · doi:10.1371/journal.pone.0231703 · Cited by 45 · see citing works
“The issue of potential navigation topology has already generated several key studies in other platforms such as Twitter or Facebook, especially with respect to polarization and fragmentation, yielding convincing graph typologies (see, inter alia, [21–25]).”
BackgroundPartisan asymmetries in exposure to misinformation
Rao, A., Morstatter, F., & Lerman, K. (2022) · Scientific Reports · doi:10.1038/s41598-022-19837-7 · Cited by 44 · see citing works
Characterizing Information Diets of Social Media Users
Kulshrestha, J., Zafar, M. B., Espín-Noboa, L., Gummadi, K. P., & Ghosh, S. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v9i1.14634 · Cited by 44 · see citing works
“There have also been some prior works on whether social media users are receiving multiple perspectives on a spe-cific event or topic (Balasubramanyan et al. 2012; Conover et al. 2011; Park et al. 2009; Adamic and Glance 2005; Borge-Holthoefer et al. 2015).”
BackgroundChipidza, W. (2021) · International Journal of Information Management · doi:10.1016/j.ijinfomgt.2021.102397 · Cited by 46 · see citing works
Aparici, R., García-Marín, D., & Rincón-Manzano, L. (2019) · El Profesional de la Informacion · doi:10.3145/epi.2019.may.13 · Cited by 43 · see citing works
Followers Are Not Enough: A Multifaceted Approach to Community Detection in Online Social Networks
Darmon, D. M., Omodei, E., & Garland, J. (2015) · PLoS ONE · doi:10.1371/journal.pone.0134860 · Cited by 42 · see citing works
“For example, [6] considered both mention and retweet networks in isolation for a collection of users chosen for their political orientation.”
BackgroundInheritance Patterns in Citation Networks Reveal Scientific MemesPreprint
Kuhn, T., Perc, M., & Helbing, D. (2014) · SSRN Electronic Journal · doi:10.2139/ssrn.2426903 · Cited by 42 · see citing works
Weismueller, J., Gruner, R. L., Harrigan, P., Coussement, K., & Wang, S. (2023) · Information Systems Journal · doi:10.1111/isj.12453 · Cited by 51 · see citing works
“…’ preference to maintain social relationships with like-minded others only, which leads to polarised networks in which there are few heterogenous social interactions on, for example, Facebook (Bakshy et al., 2015; Kitchens et al., 2020) and Twitter (Boutyline & Willer, 2017; Conover et al., 2011).”
BackgroundPredicting Twitter User Demographics from Names Alone
Wood-Doughty, Z., Andrews, N., Marvin, R., & Dredze, M. H. (2018) · doi:10.18653/v1/w18-1114 · Cited by 41 · see citing works
“…relationships in social networks can also be informative (Chen et al., 2015; Volkova et al., 2014; Bergsma et al., 2013); people tend to be friends with people who live in the same geographic area (Jurgens, 2013) or tend to follow users with similar political orientations (Conover et al., 2011).”
BackgroundRetweet networks of the European Parliament: evaluation of the community structure
Cherepnalkoski, D., & Mozetič, I. (2016) · Applied Network Science · doi:10.1007/s41109-016-0001-4 · Cited by 40 · see citing works
“Conover et al. 2011a predict the political alignment of Twitter users in the run-up to the 2010 US elections based on content and network structure.”
BackgroundThe systemic impact of deplatforming on social media
Mekacher, A., Falkenberg, M., & Baronchelli, A. (2023) · PNAS Nexus · doi:10.1093/pnasnexus/pgad346 · Cited by 48 · see citing works
Bliuc, A., Smith, L. G. E., & Moynihan, T. (2020) · Group Processes & Intergroup Relations · doi:10.1177/1368430220942567 · Cited by 40 · see citing works
Network segregation and the propagation of misinformation
Stein, J., Keuschnigg, M., & van de Rijt, A. (2023) · Scientific Reports · doi:10.1038/s41598-022-26913-5 · Cited by 38 · see citing works
The anatomy of a population-scale social network
Bokányi, E., Heemskerk, E. M., & Takes, F. W. (2023) · Scientific Reports · doi:10.1038/s41598-023-36324-9 · Cited by 39 · see citing works
Unsupervised Model for Topic Viewpoint Discovery in Online Debates Leveraging Author Interactions
Trabelsi, A., & Zai͏̈ane, O. R. (2018) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v12i1.15021 · Cited by 38 · see citing works
Political polarization in social media: Analysis of the “Twitter political field” in Japan
Takikawa, H., & Nagayoshi, K. (2017) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata.2017.8258291 · Cited by 40 · see citing works
“” Conover and colleagues hand-coded the tweets as liberal or conservative and found one-sided argument (either liberal or conservative) in each community in RT network [9].”
BackgroundInfluentialJiang, J., Ren, X., & Ferrara, E. (2023) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v17i1.22160 · Cited by 39 · see citing works
Arabic Offensive Language Classification on Twitter
Mubarak, H., & Darwish, K. M. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-34971-4_18 · Cited by 37 · see citing works
A multi-source integration framework for user occupation inference in social media systems
Huang, Y., Yu, L., Wang, X., & Cui, B. (2014) · World Wide Web · doi:10.1007/s11280-014-0300-6 · Cited by 35 · see citing works
“As pointed out by [7], the mention function (@) in social network, often plays a role to connect users having different or even opposed opinions and behaviors, which results in the poor homophily phenomenon in mention network and low inference performance consequently.”
Background
- Authors
- Conover, M., Ratkiewicz, J., Francisco, M., Goncalves, B., Menczer, F., & Flammini, A.
- Affiliations
- Indiana University (US)
- Venue
- Proceedings of the International AAAI Conference on Web and Social Media
- Published
- 2021-08-03
- Volume
- 5
- Issue
- 1
- Pages
- 89-96
- ISSN
- 2162-3449 (ISSN-L)
- 2334-0770
- OpenAlex
- W4312820073
- Semantic Scholar
- 9a7ae269358c6366cc46dc924d63b47ba405d082
- Cited by
- 626
- References
- 21
- Citations by year
- 2026: 34 · 2025: 66 · 2024: 86 · 2023: 82 · 2022: 65 · 2021: 62 · 2020: 45 · 2019: 33 · 2018: 40 · 2017: 26 · 2016: 22 · 2015: 25 · 2014: 21 · 2013: 10 · 2012: 7
- FWCI
- 107.66
- Type
- conference paper
- Language
- en
- Open access
- diamond
- doi.org
- Locations
- Proceedings of the International AAAI Conference on Web and Social Media (publishedVersion)
- AAAI Publications (The Association for the Advancement of Artificial Intelligence (AAAI)) (submittedVersion)
- Topics
- Social Sciences › Communication › Social Media and Politics
- Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
- Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Complex Network Analysis Techniques
- Keywords
- political polarization, Twitter, retweets, mentions, social network topology, 2010 midterm elections, political communication, information dissemination, network clustering
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-13.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-13.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
