Do Search Algorithms Endanger Democracy? An Experimental Investigation of Algorithm Effects on Political Polarization
Open accessAbstract
This study examines algorithm effects on user opinion, utilizing a real-world recommender algorithm of a highly popular video-sharing platform, YouTube. We experimentally manipulate user search/watch history by our custom programming. A controlled laboratory experiment is then conducted to examine whether exposure to algorithmically recommended content reinforces and polarizes political opinions. Results suggest that political self-reinforcement, as indicated by the political emotion-ideology alignment, and affective polarization are heightened by political videos – selected by the YouTube recommender algorithm – based on participants’ own search preferences. Suggestions for how to reduce algorithm-induced political polarization and implications of algorithmic personalization for democracy are discussed.
Verified 2026-10-11 · 1 reference not checked yet
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
7 of 54 cited sources checked · 2 supported · 0 flagged · accuracy not yet measured
- Premisecitation
Existing evidence suggests that the pattern of individuals' information seeking and consumption is more nuanced and multifaceted than assumed in the notion of selective exposure and homophily.
“Existing evidence suggests that the pattern of individuals' information seeking and consumption is more nuanced and multifaceted than assumed in the notion of selective exposure and homophily.”
Quote in the paperSubstantial claim1 of 3 sources checked- cites Beam, 2014 source
- cites Garrett et al., 2013 source
- cites Hart et al., 2009 source
The Results (page unavailable) show that information selection depends on utility: participants sometimes favored information that challenged their existing views. This supports a more nuanced account of selective exposure, but the paper does not assess homophily, so that part of the comparison is unsupported. (LLM, conf=0.82)
Experimental — accuracy not yet measured
- Findingown data
Affective polarization is heightened by political videos selected by the YouTube recommender algorithm based on participants' own search preferences.
“Results suggest that political self-reinforcement, as indicated by the political emotion-ideology alignment, and affective polarization are heightened by political videosselected by the YouTube recommender algorithmbased on participants' own search preferences.”
Quote in the paperSubstantial claim - Findingown data
Political self-reinforcement, indicated by political emotion-ideology alignment, is heightened by political videos selected by the YouTube recommender algorithm based on participants' own search preferences.
“Results suggest that political self-reinforcement, as indicated by the political emotion-ideology alignment, and affective polarization are heightened by political videosselected by the YouTube recommender algorithmbased on participants' own search preferences.”
Quote in the paperSubstantial claim - Findingown data
Ideas perceived to be discussed in one's social circles, if adopted as search terms by the user, weaken reinforcing effects of algorithms.
“As shown in the results, ideas perceived to be discussed in one's social circles, if adopted as search terms by the user, weaken reinforcing effects of algorithms.”
Quote in the paperSubstantial claim - Findingown data
The pattern of stronger ideology-emotion alignment in the 'self' condition than in the 'social' condition is particularly pronounced for negative emotions such as anger and sadness.
“This pattern is particularly pronounced for negative emotions such as anger and sadness.”
Quote in the paperNot a substantial claim
- Findingown data
There were no significant differences in affective polarization across the three conditions (self, social, control), especially for specific discrete emotions (anger, fear, sadness, joy, hope).
“The results suggest that there were no significant differences in affective polarization across the three conditions (i.e., self, social, and control), especially when it came to specific discrete emotions (i.e., anger, fear, sadness, joy, and hope) (see Table 2).”
Quote in the paperSubstantial claim - Findingown data
The difference between the social condition and the control condition for sadness is statistically significant based on z-statistics.
“Of these, the difference between the social condition and the control condition for sadness turned out to be statistically significant, based on z-statistics.”
Quote in the paperNot a substantial claim - Findingreasoning
The negative correlation between relative anger and political ideology across all conditions indicates that participants who identified as liberal tend to feel less anger toward Clinton than toward Trump.
“the negative correlation across all conditions indicates, not surprisingly, that participants who identified themselves as liberal tend to feel less anger toward Clinton than toward Trump.”
Quote in the paperNot a substantial claim - Findingown data
The pattern of reinforcement and polarization was less salient when algorithms were fed with socially generated search terms.
“In contrast, such a pattern of reinforcement and polarization was less salient when algorithms were fed with socially generated search terms.”
Quote in the paperSubstantial claim - Findingown data
The results of our study suggest that the reinforcing effects of algorithms diminish as users employ search terms cued by social networks.
“the results of our study suggest that the reinforcing effects of algorithms diminish as users employ search terms cued by social networks”
Quote in the paperSubstantial claim - Interpretationreasoning
Tailoring search outcomes to reinforcing information is key to user engagement and satisfaction and thus to a platform's economic success.
“This means that tailoring search outcomes to reinforcing information is key to user engagement and satisfaction and thus to a platform's economic success.”
Quote in the paperNot a substantial claim - Interpretationreasoning
The interactive search process between users and algorithms likely serves to reinforce users' prior preference structure, because recommending algorithms are responsive to users' past behaviors and search outcomes are reflective of user interest and preferences.
“When viewed in this light, the interactive search process between users and algorithms likely serves to reinforce users' prior preference structure, since recommending algorithms are responsive to users' past behaviors and thus search outcomes are reflective of user interest and preferences.”
Quote in the paperSubstantial claim - Interpretationreasoning
Users' initial pattern of search term selection and information consumption, largely driven by prior attitudes and preferences, acquaints algorithms with users and allows for more accuracy in personalized recommendations.
“Users' initial pattern of search term selection and information consumptionwhich is largely driven by prior attitudes and preferencesacquaints the algorithms with the users and thus allows for more accuracy in personalized recommendations.”
Quote in the paperNot a substantial claim - Premisecitation
Little is known about the social and political implications of search algorithms.
“Despite these increasing concerns and debates, little is known about the social and political implications of search algorithms (Hilbert et al., 2018;Thorson & Wells, 2015).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Hilbert et al., 2018 source
- cites Thorson & Wells, 2015
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationreasoning
Selective exposure favoring information that confirms one's existing beliefs would be further amplified if individuals' information search and consumption are aided by recommender algorithms.
“Selective exposure favoring information that confirms one's existing beliefs would be further amplified if individuals' information search and consumption are aided by recommender algorithms.”
Quote in the paperNot a substantial claim - Interpretationcitation
Given that ideology serves as a foundation of political affect, confirmation of existing ideology would likely result in stronger affect.
“Given that ideology serves as a foundation of political affect (Fiske & Pavelchak, 1986), the confirmation of existing ideology would likely result in stronger affect.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Fiske & Pavelchak, 1986
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationcitation
Selective exposure to like-minded information would likely result in affective polarization, a process in which individuals develop diverging evaluations of attitude objects in a partisan context.
“Second, selective exposure to like-minded information would also likely result in affective polarizationa process of opinion formation in which individuals develop diverging evaluations of attitude objects (e.g., in-group vs. out-group candidates) in a partisan context (Abramowitz & Webster, 2016;Iyengar et al., 2012;Levendusky, 2018).”
Quote in the paperSubstantial claim0 of 3 sources checked - Interpretationreasoning
Algorithmic personalization based on user preference for attitude-consonant information would structurally limit the opportunity to encounter attitude-dissonant information that potentially challenges existing affective attachment to the in-group or antipathy toward the out-group.
“Yet, algorithmic personalization based on user preference for attitude-consonant information would structurally limit the opportunity to encounter attitude-dissonant information that potentially challenges existing affective attachment to the in-group and/or antipathy toward the out-group.”
Quote in the paperSubstantial claim - Premisecitation
Ideas available in social networks, especially those based on online platforms, could go beyond the boundary of self and be more diverse than self-generated ideas.
“Although social network construction is self-selective, ideas available in social networks, especially those based on online platforms, could go beyond the boundary of self and be more diverse than selfgenerated ideas (Kwak et al., 2010;Song et al., 2020).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Kwak et al., 2010 source
- cites Song et al., 2020
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisereasoning
Pro-attitudinal information is easier to comprehend and accept because it does not create cognitive dissonance or provoke extra cognitive processing.
“Pro-attitudinal information is easier to comprehend and accept as it does not create cognitive dissonance, nor provoke extra cognitive processing.”
Quote in the paperNot a substantial claim - Premisecitation
There is an increasing tendency that people, especially partisans, defend the in-group and dislike the out-group.
“Recent work on public opinion suggests that there is an increasing tendency that people, especially partisans, defend the in-group and dislike the out-group (Iyengar et al., 2012;Levendusky, 2018).”
Quote in the paperSubstantial claim0 of 2 sources checked - Interpretationreasoning
If the findings from YouTube can be generalized to search engines and algorithms running on other online platforms, algorithms and big data are responsible, at least in part, for the increasingly divisive political culture in many contemporary democracies.
“If our findings from YouTube can be generalized to search engines and algorithms running on other online platforms, algorithms and big data are responsible, at least in part, for the increasingly divisive political culture in many contemporary democracies.”
Quote in the paperSubstantial claim - Interpretationown data
There is the possibility that political selectivity can be lessened by algorithms if a range of search terms that goes beyond users' political beliefs and identity is used.
“Thus, there is the possibility that political selectivity can be lessened by algorithms if a range of search terms that goes beyond users' political beliefs and identity is used.”
Quote in the paperSubstantial claim - Premisecitation
In social media, individuals often consume news from ideologically misaligned sources when it is recommended by their friends.
“Further, in the context of social media, Messing and Westwood (2014) found that individuals often consume news from ideologically misaligned sources when it is recommended by their friends.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Messing and Westwood (2014) source
- Interpretationreasoning
Key to the nature of algorithm effects is how people search and consume information.
“Given the reinforcing dynamics between user inputs and algorithmic personalization, key to the nature of algorithm effects is how people search and consume information.”
Quote in the paperSubstantial claim - Interpretationreasoning
The tendency toward affective polarization would likely be magnified when people are exposed to like-minded information, as it endorses their existing partisan attachment and evaluation.
“This tendency toward affective polarization would likely be magnified when people are exposed to like-minded information, as it endorses their existing partisan attachment and evaluation.”
Quote in the paperSubstantial claim - Premisecitation
Political affect is in large part rooted in one's long-standing political predispositions, such as ideology.
“That is, political affectas”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Marcus, 2000 source
- Interpretationown data
Affective polarization is not heightened by videos YouTube recommends based on either users' self-preferences or preferences of users' social networks, as affective polarization in both conditions is no greater than that in the control condition.
“the results indicate that affective polarization is not heightened by videos YouTube recommendsbased on either users' self-preferences or preferences of users' social networksas affective polarization in both conditions is no greater than that in the control condition.”
Quote in the paperSubstantial claim - Study design and conductown data
Including age as a covariate in the analyses produced substantially the same results as those reported.
“When we included age in our analyses as a covariate, the results were substantially the same as those we reported below.”
Quote in the paperNot a substantial claim - Study design and conductown data
Every relevant preexposure variable (discrete emotions, feeling thermometer measures, ideology, political interest, sex, race, college grade level) was randomly distributed across conditions, with the exception of age.
“The results indicate that every relevant preexposure variable (i.e., discrete emotions, feeling thermometer measures, ideology, political interest, sex, race, college grade level) is randomly distributed across conditions with the exception of age.”
Quote in the paperSubstantial claim - Premisecitation
Individuals tend to have positive feelings toward an object on their ideological side while attaching negative affective tags to that on the opposite side.
“Individuals tend to have positive feelings toward an object on their ideological side while attaching negative affective”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Fiske & Pavelchak, 1986
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationreasoning
The structure of the affect-ideology link would likely be reinforced when individuals are exposed to proattitudinal information.
“The structure of affect-ideology link would then likely be reinforced when individuals are exposed to proattitudinal information.”
Quote in the paperSubstantial claim - Interpretationreasoning
Through a positive feedback loop between users and recommender algorithms, selective exposure, starting from directional or defensive motivation, will be structurally facilitated by invisible recommender systems.
“Through this positive feedback loop between the users and recommender algorithms, selective exposurewhich starts off based on users' directional or defensive motivationwill be structurally facilitated by the invisible recommender systems.”
Quote in the paperNot a substantial claim - Interpretationreasoning
Exposure to attitude-consistent information corroborates existing opinions and attitudes, making them more rigid and extreme.
“Exposure to such information then easily corroborates existing opinions and attitudes, making them more rigid and extreme.”
Quote in the paperSubstantial claim - Premisecitation
Social endorsement trumps partisan selectivity in shaping individuals' choice of what information to consume.
“In addition to the internal motivation toward accuracy and understanding, social endorsement trumps partisan selectivity in shaping individuals' choice of what information to consume (Messing & Westwood, 2014).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Messing & Westwood, 2014 source
- May combine two claims. 24
- Classified as probably not a substantial claim. 40
- The quote only partly states this claim. 2
May combine two claims.
PremisecitationAlthough people do not necessarily avoid attitude-discrepant messages, attitude-consistent messages that resonate with them and confirm their desired conclusions are more likely to be preferred.
“Although people do not necessarily avoid attitude-discrepant messages (Ahn et al., 2014;Beam, 2014;Garrett et al., 2013), it is more likely that attitude-consistent messages that resonate with them and confirm their desired conclusions are preferred (Feldman et al., 2018;Gentzkow & Shapiro, 2011;Stroud, 2011).”
Quote in the paperSubstantial claim1 of 3 sources checked- cites Feldman et al., 2018 source
- cites Gentzkow & Shapiro, 2011 source
The Introduction reports that conservatives and liberals have somewhat different ideological exposure in their online news diets, consistent with some selection of attitude-consistent outlets. But Section 6 says the study cannot directly observe which stories people choose within those outlets, and the Introduction says its evidence does not address how people translate content into beliefs. It therefore does not establish that messages are preferred because they resonate with readers or confirm desired conclusions. (LLM, conf=0.85)
Experimental — accuracy not yet measured - cites Stroud, 2011 source
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationRecommender algorithms installed on online platforms help users efficiently access the information they are looking for while also reducing information overload.
“Recommender algorithms installed on online platforms help users efficiently access the information they are looking for while also reducing information overload (Hannak et al., 2013;Lazer, 2015).”
Quote in the paperNot a substantial claim1 of 2 sources checked- cites Hannak et al., 2013 source
The Introduction describes personalized search as ranking webpages a user is likely to be interested in more highly, which supports the information-finding part of the claim in principle. But the Recommendations section says the study did not determine whether personalization improved users’ search experiences, and the passages do not establish that it reduces information overload. (LLM, conf=0.83)
Experimental — accuracy not yet measured - cites Lazer, 2015 source
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationBased on some kind of input, recommender systems select a set of items from a large pool of available items and recommend a list of information links for users to choose.
“Based on some kind of input, recommender systems select a set of items from a large pool of available items and recommend a list of information links for users to choose (Resnick & Varian, 1997).”
Quote in the paperNot a substantial claim1 of 1 sources checked- cites Resnick & Varian, 1997 source
The Recommender Systems section says systems take recommendations as inputs, aggregate them, and direct them to recipients; it also describes recommending interesting items. But it does not characterize recommender systems generally as selecting from a large pool and presenting users with a list of information links. It describes links as one possible type of recommendation alongside books, movies, and other items. (LLM, conf=0.84)
Experimental — accuracy not yet measured
Classified as probably not a substantial claim.
PremisecitationThere are two distinct but coexisting human motivations that account for citizens' choice and consumption of political information: directional versus accuracy.
“It is proposed that there are two distinct but coexisting human motivations that account for citizens' choice and consumption of political information: directional vs. accuracy (Kunda, 1990, or defensive vs. accuracy in Hart et al., 2009, or cognitive consistency vs. understanding in Holbert et al., 2013).”
Quote in the paperSubstantial claim1 of 3 sources checkedClassified as probably not a substantial claim.
Study design and conductcitationIdeological reinforcement was examined by the correlation between affective candidate evaluations and political ideology, which quantifies the extent to which the two are aligned as a function of experimental treatment.
“First, ideological reinforcement was examined by the correlation between affective candidate evaluations and political ideology which quantifies the extent to which the two are aligned as a function of the experimental treatment (Cho et al., 2018).”
Quote in the paperSubstantial claim1 of 1 sources checked- cites Cho et al., 2018 source
The Results describe testing whether political expression strengthens the relationship between party identification and political opinions, and the Measures section describes affective ratings of candidates (page numbers unavailable). But the paper tests that relationship with regression models and tracks changes in candidate preference; it does not describe correlating candidate evaluations with ideology as a function of experimental treatment. (LLM, conf=0.86)
Experimental — accuracy not yet measured
May combine two claims.
PremisecitationPeople generally do not necessarily avoid counter-attitudinal information, and further seek such information when they believe it helps them understand and be accurate on the situation at hand.
“People, in general, do not necessarily avoid counter-attitudinal information (Beam, 2014;Garrett et al., 2013), and further seek such information when they believe it helps them understand and be accurate on the situation at hand (Hart et al., 2009;Holbert et al., 2013;Kunda, 1990).”
Quote in the paperSubstantial claim1 of 3 sources checkedProbably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataIn the ANCOVA model, political interest added as a random factor interacting with condition was not significant.
“in the ANCOVA model, political interest was added as a random factor and specified to interact with the fixed factor (i.e., conditions), and the results for the interaction were not significant.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims. The quote only partly states this claim.
Findingown dataFor fear, joy, hope, and overall valence (feeling thermometer), the ideology-emotion alignment was stronger in the 'self' condition than in the 'social' condition, but these differences were not statistically significant.
“A similar, albeit not statistically significant, the pattern was observed for all other measures of emotion.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataThe three-way interaction (ideology x experimental condition x interest) predicting post-exposure candidate evaluation, with pre-exposure candidate evaluation controlled, was not significant; likewise, two-way interactions involving interest were not significant.
“The three-way interaction was not significant; likewise, two-way interactions involving interest were not significant either.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataThe experiment showed an overall pattern of opinion reinforcement and polarization after exposure to algorithm-recommended content, especially when algorithms were fed with self-generated search terms.
“The results of our experiment showed an overall pattern of opinion reinforcement and polarization after exposure to algorithm-recommended content, especially when algorithms were fed with self-generated search terms.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationWith accuracy motivation, individuals often cross ideological lines and glean, or at least do not avoid, information from the other side of the political spectrum.
“With this motivation, individuals often cross ideological lines and glean (or at least do not avoid) information from the other side of the political spectrum (Garrett et al., 2013).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Garrett et al., 2013 source
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationThe accuracy motivation drives individuals to seek information they expect to be of high informational value.
“In contrast, the accuracy motivation drives individuals to seek information that they expect to be of high informational value.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Garrett et al., 2013 source
May combine two claims.
PremisecitationAlgorithms are naturally designed to operate with, not against, human tendencies, one of which is that people feel more comfortable with consistency and confirmation.
“The underlying reason for the reinforcing tendency of algorithms is that algorithms are naturally designed to operate with, not against, human tendencies, one of which is that people feel more comfortable with consistency and confirmation (Kunda, 1990).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Kunda, 1990 source
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationThe input to recommender systems can be a search term or some other preference statement, and selection is traditionally based on the personal history of the user (content-based filtering), the cross-section of other users (collaborative filtering), or a mix of both.
“The input can be in the form of a search term or some other preference statement, and the selection is traditionally based on the personal history of the user (content-based filtering), or the cross-section of other users (collaborative filtering), or a mix of both (Balabanović & Shoham, 1997).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Balabanović & Shoham, 1997 source
Classified as probably not a substantial claim.
InterpretationreasoningAs data about a user's information search and consumption accumulates, algorithms' ability to personalize is sharpened in subsequent searches.
“As data about information search and consumption for a specific user accumulates, algorithms' ability to personalize is sharpened in subsequent searches.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S4 (based).
InterpretationreasoningUser-oriented recommendations increase the likelihood of consumption (clicking) by users because the recommended messages fit their interests.
“Such user-oriented recommendations, in turn, increase the likelihood of consumption (clicking) by the users, as the recommended messages fit their interests.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
PremisereasoningTo the extent that user inputs reflect existing political preferences, search engine use could eventually lead to a filter bubble: a like-minded information environment created by personalized algorithmic recommendations and resulting segregation of users in their ideological and political views.
“To the extent that user inputs (e.g., search terms, clicks, etc.) reflect their existing political preferences, search engine use could eventually lead to the so-called”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Pariser, 2011
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Classified as probably not a substantial claim.
PremisecitationThe directional (or defensive) motivation leads to selective exposure, encouraging individuals to seek information or news sources congenial to their existing attitudes.
“The directional (or defensive) motivation leads to selective exposure, encouraging individuals to seek information (or news sources) that they believe is congenial to their existing attitudes (Iyengar & Hahn, 2009;Stroud, 2011).”
Quote in the paperSubstantial claim0 of 2 sources checkedClassified as probably not a substantial claim.
Study design and conductnone givenAffective polarization was measured as the difference between affective evaluations of the two candidates, which captures the degree to which attitudes about in-group and out-group candidates diverge.
“Second, affective polarization was measured as the difference between affective evaluations of the two candidates which captures the degree to which attitudes about in-group and out-group candidates diverge.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationown dataAlgorithmic recommendations personalized by user behavior data, if unchecked, have the potential to solidify personal political convictions and encourage polarized opinions.
“Thus, algorithmic recommendations personalized by user behavior data, if unchecked, has the potential to solidify personal political convictions and encourage polarized opinions.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductreasoningThe control condition videos were a baseline control because they were what a user searching the same topic without a specific search/watch history would encounter per YouTube's recommendation at the time.
“These videos were used as a baseline controlsince they were the videos that a user who was in search for videos about the same topic (i.e., 410 the presidential election of past week)but did not have a specific search/watch history would encounter, per YouTube's recommendation at the time.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
PremisecitationExposure to information that is more diverse or less agreeable in relation to one's own thoughts and beliefs could lessen self-reinforcement and opinion polarization.
“Exposure to information that is more diverse or less agreeable in relation to one's own thoughts and beliefs could then lessen selfreinforcement and opinion polarization (Mutz, 2006).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Mutz, 2006 source
Probably not a substantial claim: may not pass S4 (based).
Interpretationnone givenNon-self-report, behavioral data about user interests and beliefs allow algorithms to hone further recommendations.
“Non-self-report, behavioral data about user interests and beliefs allow algorithms to hone further recommendations.”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The terms that users enter for their online searches are reflective of what they care about and what they believe.
May combine two claims.
PremisecitationAttitude-consistent information, or information from a like-minded source, is often viewed as more credible and of higher quality, and this perception encourages consumption and facilitates processing.
“Further, attitude-consistent information (or information from a like-minded source) is often viewed as more credible and of higher quality, and this perception not only encourages consumption of the information but also facilitates the processing of it (Miller et al., 1993).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Miller et al., 1993 source
May combine two claims.
Interpretationown dataDue to programmed responsiveness to past user interests and preferences, algorithms serve as a confirmatory communication partner or an echo chamber that reassures and reinforces users' prior beliefs and fosters extremism, as observed in politically homogeneous communication networks.
“Due to the programmed responsiveness to past user interests and preferences, algorithms serve as a confirmatory communication partner or an echo chamber that reassures and reinforces users' prior beliefs and fosters extremism, as observed in the case of politically homogeneous communication networks.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim. May combine two claims.
InterpretationreasoningExposure to like-minded information via personalized algorithmic recommendations shapes users' political opinion in two ways: ideological reinforcement and affective polarization.
“When viewed from the broader literature in political psychology, this research on selective exposure suggests two ways how exposure to likeminded information via personalized algorithmic recommendations would shape users' political opinion: ideological reinforcement and affective polarization.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationown dataIdeological reinforcement, as indicated by political emotion-ideology alignment, is heightened by political videos selected by the YouTube recommender algorithm based on participants' own search preferences and is muted to some degree if perceived preferences of participants' online social circle are used as inputs.
“This suggests that ideological reinforcement, as indicated by the political emotion-ideology alignment, is heightened by political videos selected by the YouTube recommender algorithm based on participants' own search preferences and is muted to some degree if perceived preferences of participants' online social circle are used as inputs.”
Quote in the paperSubstantial claim- Rests on an assumption: Personalization by users' past behavior is a key mechanism of online search and recommender algorithms.
- Rests on an assumption: Algorithmic recommendations are a product of users' search terms and subsequent consumption of information.
Probably not a substantial claim: may not pass S4 (based). May combine two claims.
InterpretationreasoningAlgorithmic recommendations are largely an outcome of iterative interactions between users and algorithms, and the recommendations differ depending on what search terms users enter.
“To summarize, algorithmic recommendations are largely an outcome of iterative interactions between users and algorithms, and the algorithmic recommendations are different depending on what search terms users enter.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
InterpretationreasoningIf a user's search term selection goes beyond her routine repertoire, algorithms would provide more diverse recommendations.
“In contrast, if a user's search term selection somehow goes beyond her routine repertoire, algorithms would provide more diverse recommendations.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationSocial endorsement reduces partisan selectivity toward like-minded information, so social media use is an opportunity to increase users' exposure to a variety of news and politically diverse information.
“As social endorsement reduces partisan selectivity toward like-minded information, social media use is an opportunity to”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Messing & Westwood, 2014 source
Classified as probably not a substantial claim.
PremisecitationHuman memory is a network of interconnected cognitive and affective constructs.
“First, schema theory in social cognition research has long suggested that human memory is a network of interconnected cognitive and affective constructs (Anderson & Bower, 1973;Lodge & Stroh, 1993).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Anderson & Bower, 1973
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- cites Lodge & Stroh, 1993 source
Probably not a substantial claim: may not pass S4 (based).
InterpretationreasoningIf individuals are motivated to explore ideas encountered in their social networks, their search behaviors would likely be less bounded by their own interests and preferences and thus train algorithms on their own.
“However, if individuals are motivated to explore ideas that they encounter in their social networks, their search behaviors would likely be less bounded by their own interests and preferences and thus train algorithms on its own.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
InterpretationcitationGiven the human tendency toward psychological consistency, cognitive salience of counter-attitudinal ideas in the user mind would not be strong enough to drive routine searches online.
“Yet, given the human tendency toward psychological consistency (Kunda, 1990), cognitive salience of counter-attitudinal ideas in the user mind would not be strong enough to drive routine searches online.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Kunda, 1990 source
May combine two claims.
InterpretationreasoningAlgorithms, although helping users navigate a massive amount of information, add to political selectivity that separates people from contradictory views.
“Viewed in this light, algorithms, although helping users navigate a massive amount of information, add to political selectivity that separates people from contradictory views.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductreasoningThe baseline is dynamic, as the current YouTube environment simultaneously affects both personalized recommendations and trending videos.
“It is a dynamic baseline, as the current YouTube environment simultaneously affects both the personalized recommendations and the trending videos.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
PremisecitationA general conclusion of past research on selective exposure is that exposure to confirmatory information strengthens existing opinions.
“A general conclusion of past research on selective exposure is that exposure to confirmatory information strengthens existing opinions (Stroud, 2011;Sunstein, 2001;Tewksbury & Riles, 2015).”
Quote in the paperSubstantial claim0 of 3 sources checkedProbably not a substantial claim: may not pass S4 (based).
InterpretationreasoningIf searches are crafted based on one's perception or memory of ideas shared in social networks, the resulting algorithmic recommendations would likely contain a wider range of information or perspectives.
“That is, if searches are crafted based on one's perception (or memory) of ideas shared in social networks, the resulting algorithmic recommendations would likely contain a wider range of information or perspectives.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
InterpretationreasoningIf information search is driven by a user's own perspectives, algorithms recommend information the user would agree with.
“If, as suggested by the notion of selective exposure, information search is driven by a user's own perspectives, algorithms recommend information the user would agree with.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningThe heightened selectivity from user-algorithm interaction induces a personally tailored information environment for each user, which in turn shapes the user's opinion.
“The heightened selectivity as a result of the user-algorithm interaction induces a personally tailored information environment for each user which, in turn, shapes the user's opinion.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
PremisecitationIndividuals' political opinion is often organized in a coherent structure of attitudes in which domain-specific opinions are grounded in broader political beliefs such as ideology.
“In light of the schema theory, scholars of public opinion have posited that individuals' political opinion is often organized in a coherent structure of attitudes in which domain-specific opinions are grounded in a broader set of political beliefs (or”
Quote in the paperSubstantial claim0 of 3 sources checkedMay combine two claims.
PremisecitationIf a person is exposed to counter-attitudinal information, her attitudinal ambivalence would increase, thereby reducing affective polarization.
“If a person is exposed to counter-attitudinal information, her attitudinal ambivalence would increase, thereby reducing affective polarization (Huckfeldt et al., 2004;Mutz, 2006).”
Quote in the paperSubstantial claim0 of 2 sources checkedClassified as probably not a substantial claim. May combine two claims. The quote only partly states this claim.
Study design and conductreasoningThe first ranking identifies self search preference while the second taps into social preference at least in participants' perception.
“The first identifies”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
InterpretationreasoningConsumption of proattitudinal information recommended by algorithms gives reinforcing feedback to algorithms, confirming users' preferences and validating and enhancing algorithmic personalization.
“Then, the consumption of proattitudinal information recommended by the algorithms gives reinforcing feedback to the algorithms, confirming the users' preferences and thus validating and enhancing algorithmic personalization.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductreasoningThe way search preferences were measured (choosing topics from a pre-selected set) only allows participants to indicate their preferences constrained within the list.
“this approach only allows participants to indicate their preferences constrained within the list.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
InterpretationreasoningThe relatively open social networks of college students likely lead to a wider diversity of search terms gleaned from their online friends, which could potentially explain the results at least in part.
“The relatively open social networks likely lead to a wider diversity of search terms gleaned from their online friends, which could potentially explain the results at least in part.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
InterpretationreasoningThe results should be interpreted as showing algorithm effects when users are somehow motivated to explore ideas that they think their friends care about.
“The results should be interpreted as showing algorithm effects when users are somehow motivated to explore ideas that they think their friends care about.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Study design and conductnone givenThis study utilized an actual real-world, not proxy, algorithm on a highly popular online platform, which greatly enhances the external validity and relevance of the findings.
“this study utilized an actual real-world, not proxy, algorithm on a highly popular online platform, a fact that greatly enhances the external validity and relevance of the findings.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationCompared to the average population, college students are in general more active on the Internet and tend to be friends with a more diverse group of people on their online social networks.
“As compared to the average population, college students are in general more active on the Internet and tend to be friends with a more diverse group of people on their online social networks (Pfeil et al., 2009).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Pfeil et al., 2009 source
Classified as probably not a substantial claim.
Study design and conductreasoningWe cannot draw a firm conclusion from a single experiment with college students.
“we cannot draw a firm conclusion from a single experiment with college students.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningCoupled with this research, our results suggest that algorithm effects on opinion reinforcement and polarization are conditional and would be better understood when considering how and to what extent cross-cutting exposure, as opposed to selective exposure, occurs.
“Coupled with this research, our results suggest that algorithm effects on opinion reinforcement and polarization are conditional and would be better understood when considering how and to what extent cross-cutting exposure, as opposed to selective exposure, occurs.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).
PremisecitationThe complex and dynamic algorithmic systems handling big data make it difficult to predict outcomes.
“the complex and dynamic algorithmic systems handling big data make it difficult to predict outcomes (Lazer, 2015).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Lazer, 2015 source
Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).
InterpretationreasoningThe unique situation of the just-finished presidential election might amplify the reinforcing effects of algorithmic personalization.
“This unique situation might amplify the reinforcing effects by algorithmic personalization as perhaps participants already have strong opinions and information recommended by algorithms might contain partisan or polarizing cues.”
Quote in the paperNot a substantial claim- Rests on an assumption: The overall body of available information online about the election and candidates was highly polarized at the time when this experiment was conducted.
- Rests on an assumption: The opinion targets (presidential candidates) are much more salient and polarizing in participants' minds than any other issues.
May combine two claims.
Interpretationown dataWhen it comes to controversial social and political topics, algorithmic personalization could also serve as an echo chamber strengthening users' prior convictions and thus aggravating the existing political chasm in society.
“When it comes to controversial social and political topics, algorithmic personalization could also serve as an echo chamber strengthening users' prior convictions and thus aggravating the existing political chasm in society.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductnone givenUser opinion in this study was operationalized as affective assessments of two major candidates of the presidential election that was just finished.
“user opinion, in this study, was operationalized as affective assessments of two major candidates of the presidential election that was just finished.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationThe first recommended video was selected based on previous evidence suggesting that users often click on top-ranked search results.
“The first recommended video was selected based on previous evidence suggesting that users often click on top-ranked search results(Bakshy et al., 2015).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Bakshy et al., 2015 source
The authors state this as an assumption.
The terms that users enter for their online searches are reflective of what they care about and what they believe.
“A key assumption is that the terms that users enter for their online searches are reflective of what they care about and what they believe.”
The paper relies on this without support.
Personalization by users' past behavior is a key mechanism of online search and recommender algorithms.
“Although personalization by users' past behavior is a key mechanism, the complex and dynamic algorithmic systems handling big data make it difficult to predict outcomes (Lazer, 2015).”
The paper relies on this without support.
Algorithmic recommendations are a product of users' search terms and subsequent consumption of information.
“since algorithmic recommendations are a product of users' search terms and subsequent consumption of information.”
The paper relies on this without support.
The overall body of available information online about the election and candidates was highly polarized at the time when this experiment was conducted.
“the overall body of available information online about the election and candidates was also likely highly polarized at the time when this experiment was conducted.”
The paper relies on this without support.
The opinion targets (presidential candidates) are much more salient and polarizing in participants' minds than any other issues.
“It is thus highly likely that the opinion targets (i.e., presidential candidates) are much more salient and polarizing in participants' minds than any other issues.”
The paper relies on this without support.
Online platforms operating in different contexts program and run their algorithms differently.
“However, online platforms operating in different contexts program and run their algorithms differently.”
The paper relies on this without support.
Online search and recommender algorithms are essentially black boxes.
“This is why online search and recommender algorithms are essentially black boxes.”
Statistics recomputed
Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.
- ConsistentF(2, 104) = 4.47reported p < .05 · recomputed p = .014
“F(2, 104) = 4.47, p < .05”
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.
Anger correlation is not negative in every condition
MinorThe paper says the anger-ideology correlation is negative across all conditions, but reports a positive correlation for the control condition.
“the negative correlation across all conditions indicates, not surprisingly, that participants who identified themselves as liberal tend to feel less anger toward Clinton than toward Trump.”
“control (r = .071,n.s.)”
Feeling-thermometer errors are labeled as standard deviations
MinorThe results text labels the values accompanying the three feeling-thermometer means as standard deviations, whereas Table 2 identifies its parenthesized values as standard errors.
“participants in the "social" condition reported a lower level of polarized affective assessments (M = 31.89,SD = 2.64) in terms of overall emotional valence (i.e., feeling thermometer) than those in the "self" condition (M = 39.87,SD = 2.74) and in the control condition (M = 42.88,SD = 2.75)”
“Entries are estimated marginal means with standard errors in parentheses.”
Cites
The case for motivated reasoning.
Kunda, Z. (1990) · Psychological Bulletin · doi:10.1037/0033-2909.108.3.480 · Cited by 9,022 · cited 6× · see citing works
“Yet, given the human tendency toward psychological consistency (Kunda, 1990), cognitive salience of counter-attitudinal ideas in the user mind would not be strong enough to drive routine searches online.”
BackgroundInfluentialWhat is Twitter, a social network or a news media?
Kwak, H., Lee, C., Park, H., & Moon, S. (2010) · Proceedings of the 19th international conference on World wide web · doi:10.1145/1772690.1772751 · Cited by 6,652 · see citing works
(2023) · doi:10.1126/science · Cited by 6,063 · see citing works
(2011) · Cited by 4,430 · see citing works
Human Inference: Strategies and Shortcomings of Social Judgment.
Cherniak, C., Nisbett, R. E., & Ross, L. D. (1983) · The Philosophical Review · doi:10.2307/2184495 · Cited by 4,196 · see citing works
Elder, J. W., Geertz, C., & Rosenhack, K. P. S. (1964) · Cited by 3,878 · see citing works
Ricci⋆, F., Kantor, P. B., Shapira, B., & Rokach, L. (2010) · Springer eBooks · doi:10.1007/978-0-387-85820-3 · Cited by 3,823 · see citing works
Resnick, P., & Varian, H. R. (1997) · Communications of the ACM · doi:10.1145/245108.245121 · Cited by 3,772 · see citing works
Exposure to ideologically diverse news and opinion on Facebook
Bakshy, E., Messing, S., & Adamic, L. A. (2015) · Science · doi:10.1126/science.aaa1160 · Cited by 3,245 · see citing works
Fab: content-based, collaborative recommendation
Balabanović, M., & Shoham, Y. (1997) · Communications of the ACM · doi:10.1145/245108.245124 · Cited by 2,934 · see citing works
Iyengar, S., Sood, G., & Lelkes, Y. (2012) · Public Opinion Quarterly · doi:10.1093/poq/nfs038 · Cited by 2,748 · cited 4× · see citing works
“Recent work on public opinion suggests that there is an increasing tendency that people, especially partisans, defend the in-group and dislike the out-group (Iyengar et al., 2012; Levendusky, 2018).”
BackgroundMethodInfluentialFilter Bubbles, Echo Chambers, and Online News Consumption
Flaxman, S., Goel, S., & Rao, J. M. (2016) · Public Opinion Quarterly · doi:10.1093/poq/nfw006 · Cited by 1,992 · see citing works
Red Media, Blue Media: Evidence of Ideological Selectivity in Media Use
Iyengar, S., & Hahn, K. S. (2009) · Journal of Communication · doi:10.1111/j.1460-2466.2008.01402.x · Cited by 1,940 · see citing works
FILTER BUBBLE: Wie wir im Internet entmündigt werden
Pariser, E. (2012) · doi:10.3139/9783446431164 · Cited by 1,849 · see citing works
Political Polarization in the American Public
Fiorina, M. P., & Abrams, S. J. (2008) · Annual Review of Political Science · doi:10.1146/annurev.polisci.11.053106.153836 · Cited by 1,798 · see citing works
Pan-cultural elements in facial displays of emotion
Ekman, P., Sorenson, E. R., & Friesen, W. V. (1969) · Science · doi:10.1126/science.164.3875.86 · Cited by 1,775 · see citing works
Hearing the Other Side: Deliberative versus Participatory Democracy
Mutz, D. C. (2006) · doi:10.1017/cbo9780511617201 · Cited by 1,687 · see citing works
Affective Intelligence and Political Judgment
Marcus, G. E., Neuman, W. R., & MacKuen, M. (2000) · Cited by 1,686 · see citing works
Hearing the Other Side: Deliberative versus Participatory Democracy
Mutz, D. C. (2006) · Choice Reviews Online · doi:10.5860/choice.44-2388 · Cited by 1,588 · see citing works
A New Era of Minimal Effects? The Changing Foundations of Political Communication
Bennett, W. L., & Iyengar, S. (2008) · Journal of Communication · doi:10.1111/j.1460-2466.2008.00410.x · Cited by 1,558 · see citing works
Frijda, N. H. (1988) · American Psychologist · doi:10.1037/0003-066x.43.5.349 · Cited by 1,533 · see citing works
Human Inference: Strategies and Shortcomings of Social Judgment.
Weiner, H. J., Nisbett, R. E., & Ross, L. D. (1981) · Journal of the American Statistical Association · doi:10.2307/2287875 · Cited by 1,468 · see citing works
(1974) · Behavior Therapy · doi:10.1016/s0005-7894(74)80111-5 · Cited by 1,429 · see citing works
Feeling validated versus being correct: A meta-analysis of selective exposure to information
Hart, W., Albarracín, D., Eagly, A. H., Brechan, I., Lindberg, M. J., & Merrill, L. (2009) · Psychological Bulletin · doi:10.1037/a0015701 · Cited by 1,292 · see citing works
Alba, J. W., & Hasher, L. (1983) · Psychological Bulletin · doi:10.1037/0033-2909.93.2.203 · Cited by 1,169 · see citing works
Cognitive, Affective and Behavioral Components of Attitudes
Rosenberg, M. J. (1960) · Medical Entomology and Zoology · Cited by 1,034 · see citing works
Ideological Segregation Online and Offline *
Gentzkow, M., & Shapiro, J. M. (2011) · The Quarterly Journal of Economics · doi:10.1093/qje/qjr044 · Cited by 980 · see citing works
The rise of negative partisanship and the nationalization of U.S. elections in the 21st century
Abramowitz, A. I., & Webster, S. (2016) · Electoral Studies · doi:10.1016/j.electstud.2015.11.001 · Cited by 978 · see citing works
Stroud, N. J. (2011) · doi:10.1093/acprof:oso/9780199755509.001.0001 · Cited by 950 · see citing works
Proceedings of the 19th international conference on World wide web
(2010) · doi:10.1145/1772690 · Cited by 907 · see citing works
Messing, S., & Westwood, S. J. (2014) · Communication Research · doi:10.1177/0093650212466406 · Cited by 902 · see citing works
How are Foreign Policy Attitudes Structured? A Hierarchical Model
Hurwitz, J., & Peffley, M. (1987) · American Political Science Review · doi:10.2307/1962580 · Cited by 807 · see citing works
Curated Flows: A Framework for Mapping Media Exposure in the Digital Age
Thorson, K., & Wells, C. (2015) · Communication Theory · doi:10.1111/comt.12087 · Cited by 764 · see citing works
Niche news: the politics of news choice
(2011) · Choice Reviews Online · doi:10.5860/choice.49-1894 · Cited by 695 · see citing works
Disagreement, Ambivalence, and Engagement: The Political Consequences of Heterogeneous Networks
Huckfeldt, R., Mendez, J. M., & Osborn, T. (2004) · Political Psychology · doi:10.1111/j.1467-9221.2004.00357.x · Cited by 576 · see citing works
Should we worry about filter bubbles?
Zuiderveen Borgesius, F. J., Trilling, D., Möller, J., Bodó, B., de Vreese, C. H., & Helberger, N. (2016) · Internet Policy Review · doi:10.14763/2016.1.401 · Cited by 541 · see citing works
Pfeil, U., Arjan, R., & Zaphiris, P. (2009) · Computers in Human Behavior · doi:10.1016/j.chb.2008.08.015 · Cited by 534 · see citing works
Americans, Not Partisans: Can Priming American National Identity Reduce Affective Polarization?
Levendusky, M. S. (2018) · The Journal of Politics · doi:10.1086/693987 · Cited by 492 · see citing works
Campaigning for hearts and minds: how emotional appeals in political ads work
(2006) · Choice Reviews Online · doi:10.5860/choice.43-5558 · Cited by 479 · see citing works
Marcus, G. E. (2000) · Annual Review of Political Science · doi:10.1146/annurev.polisci.3.1.221 · Cited by 468 · see citing works
Campaigning for Hearts and Minds: How Emotional Appeals in Political Ads Work
Sears, D. O. (2006) · Perspectives on Politics · doi:10.1017/s1537592706400475 · Cited by 428 · see citing works
Measuring personalization of web search
Hannak, A., Sapiezynski, P., Molavi Kakhki, A., Krishnamurthy, B., Lazer, D., Mislove, A., & Wilson, C. (2013) · Proceedings of the 22nd international conference on World Wide Web · doi:10.1145/2488388.2488435 · Cited by 396 · see citing works
How People Organize the Political World: A Schematic Model
Conover, P. J., & Feldman, S. (1984) · American Journal of Political Science · doi:10.2307/2110789 · Cited by 378 · see citing works
Attitude organization and the attitude-behavior relationship.
Bagozzi, R. P., & Burnkrant, R. E. (1979) · Journal of Personality and Social Psychology · doi:10.1037/0022-3514.37.6.913 · Cited by 363 · see citing works
Us Against Them: Ethnocentric Foundations of American Opinion
Kinder, D. R., & Kam, C. D. (2009) · doi:10.7208/chicago/9780226435725.001.0001 · Cited by 349 · see citing works
A Turn Toward Avoidance? Selective Exposure to Online Political Information, 2004-2008
Garrett, R. K., Carnahan, D., & Lynch, E. K. (2013) · Political Behavior · doi:10.1007/s11109-011-9185-6 · Cited by 311 · see citing works
Partisan Paths to Exposure Diversity: Differences in Pro- and Counterattitudinal News Consumption
Garrett, R. K., & Stroud, N. J. (2014) · Journal of Communication · doi:10.1111/jcom.12105 · Cited by 308 · see citing works
The Duality of Media: A Structurational Theory of Public Attention
Webster, J. G. (2011) · Communication Theory · doi:10.1111/j.1468-2885.2010.01375.x · Cited by 220 · see citing works
Campaigning for Hearts and Minds
Brader, T. A. (2005) · University of Chicago Press eBooks · doi:10.7208/chicago/9780226788302.001.0001 · Cited by 207 · see citing works
Miller, A. G., McHoskey, J. W., Bane, C. M., & Dowd, T. G. (1993) · Journal of Personality and Social Psychology · doi:10.1037/0022-3514.64.4.561 · Cited by 188 · see citing works
Automating the News: How Personalized News Recommender System Design Choices Impact News Reception
Beam, M. A. (2014) · Communication Research · doi:10.1177/0093650213497979 · Cited by 159 · see citing works
The rise of the social algorithm
Lazer, D. (2015) · Science · doi:10.1126/science.aab1422 · Cited by 133 · see citing works
Influencing Myself: Self-Reinforcement Through Online Political Expression
Cho, J., Ahmed, S., Keum, H., Choi, Y. J., & Lee, J. H. (2018) · Communication Research · doi:10.1177/0093650216644020 · Cited by 116 · see citing works
Self-citationIdeological Segregation Online and OfflinePreprint
Gentzkow, M. A., & Shapiro, J. M. (2010) · SSRN Electronic Journal · doi:10.2139/ssrn.1588920 · Cited by 105 · see citing works
Inside the Mental Voting Booth: An Impression-Driven Process Model of Candidate Evaluation
Lodge, M., & Stroh, P. (1993) · Explorations in Political Psychology · doi:10.1215/9780822396697-011 · Cited by 66 · see citing works
Hilbert, M., Ahmed, S., Cho, J., Liu, B., & Luu, J. (2018) · Communication Methods and Measures · doi:10.1080/19312458.2018.1479843 · Cited by 49 · see citing works
Self-citationMotivated Reasoning, Affect, and the Role of Memory in Voter Decision Making
Redlawsk, D. P. (2006) · Palgrave Macmillan US eBooks · doi:10.1057/9781403983114_6 · Cited by 48 · see citing works
Song, H., Cho, J., & Benefield, G. A. (2018) · Communication Research · doi:10.1177/0093650218790144 · Cited by 47 · see citing works
Self-citationPolarization as a Function of Citizen Predispositions and Exposure to News on the Internet
Tewksbury, D., & Riles, J. M. (2015) · Journal of Broadcasting & Electronic Media · doi:10.1080/08838151.2015.1054996 · Cited by 42 · see citing works
Feldman, L., Wojcieszak, M., Stroud, N. J., & Bimber, B. (2018) · Journal of Broadcasting & Electronic Media · doi:10.1080/08838151.2017.1375502 · Cited by 25 · see citing works
Experts, Activists, and Interdependent Citizens: Are Electorates Self-Educating?
Ahn, T., Huckfeldt, R., & Ryan, J. (2014) · Cambridge University Press eBooks · doi:10.1017/cbo9781107706156 · Cited by 21 · see citing works
Holbert, R. L., Weeks, B. E., & Esralew, S. (2013) · American Behavioral Scientist · doi:10.1177/0002764213490693 · Cited by 16 · see citing works
Flagged first, then by citations
Research works printed but not linked: 4
- Fiske S. T., 1986, Handbook of motivation and cognition: Foundations of social behavior, 167
- Rosenberg M. J., 1960, Attitude organization and change, 1
- Sunstein C. R., 2001, Echo chambers: Bush v. Gore, impeachment, and beyond
- Zuiderveen Borghesius F., 2016, Internet Policy Review, 5, 1
Cited by
Sorted by citations
The role of (social) media in political polarization: A systematic review
Kubin, E., & von Sikorski, C. (2021) · Annals of the International Communication Association · doi:10.1080/23808985.2021.1976070 · Cited by 610 · see citing works
“Researchers found that YouTube algorithm recommendations (Cho et al., 2020), and exposure to social media comments that derogate political adversaries (Suhay et al., 2018) increases affective polarization.”
BackgroundExamining the consumption of radical content on YouTube
Hosseinmardi, H., Ghasemian, A., Clauset, A., Mobius, M., Rothschild, D. M., & Watts, D. J. (2021) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2101967118 · Cited by 171 · see citing works
Misunderstanding the harms of online misinformation
Budak, C., Nyhan, B., Rothschild, D. M., Thorson, E., & Watts, D. J. (2024) · Nature · doi:10.1038/s41586-024-07417-w · Cited by 156 · see citing works
“…how social media platforms can recommend problematic content 92 and controlled experiments suggest algorithmic recommendations can alter the relative prevalence of conservative versus liberal information 93 , limit exposure to counter-attitudinal news 94 and have negative attitudinal effects 95 .”
Shin, D., Koerber, A., & Lim, J. S. (2025) · New Media & Society · doi:10.1177/14614448241234040 · Cited by 137 · see citing works
“…can fill the gap in research on user behaviors related to misinformation that previous studies have attempted to elucidate (e.g. Ahmed and Gil-Lopez, 2022; Cho et al., 2020; Hwang and Jeong, 2023; Kim et al., 2020; Melchior and Oliveira, 2022; Molina and Sundar, 2023; Tully et al., 2020).”
Eg, R., Tønnesen, O. D., & Tennfjord, M. K. (2022) · Computers in Human Behavior Reports · doi:10.1016/j.chbr.2022.100253 · Cited by 137 · see citing works
Van Bavel, J. J., Robertson, C. E., del Rosario, K., Rasmussen, J., & Rathje, S. (2024) · Annual Review of Psychology · doi:10.1146/annurev-psych-022123-110258 · Cited by 121 · see citing works
Pedreschi, D., Pappalardo, L., Ferragina, E., Baeza-Yates, R., Barabási, A.-L., Dignum, F., Dignum, V., Eliassi-Rad, T., Giannotti, F., Kertész, J., et al. (2025) · Artificial Intelligence · doi:10.1016/j.artint.2024.104244 · Cited by 102 · see citing works
“Cho et al. [92] analysed factors leading to political polarisation on YouTube, finding that a recommender can contribute to this polarisation, but mainly through the channel of a user’s preference.”
AI Literacy in K-12 and Higher Education in the Wake of Generative AI: An Integrative Review
Gu, X., & Ericson, B. J. (2025) · arXiv · doi:10.48550/arxiv.2503.00079 · Cited by 73
A Literature Review of Video-Sharing Platform Research in HCI
Bartolome, A., & Niu, S. (2023) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581107 · Cited by 66 · see citing works
AI Literacy in K-12 and Higher Education in the Wake of Generative AI: An Integrative Review
Gu, X., & Ericson, B. J. (2025) · ACM Conference on International Computing Education Research (ICER) · doi:10.1145/3702652.3744217 · Cited by 113 · see citing works
Digital Domination: Social Media and Contestatory Democracy
Aytaç, U. (2022) · Political Studies · doi:10.1177/00323217221096564 · Cited by 121 · see citing works
“Recent lab experiments confirm this insight by showing how algorithms can reinforce and polarize political opinions by iterating users’ past preferences (Cho et al., 2020).”
BackgroundNews recommender systems: a programmatic research review
Mitova, E., Blassnig, S., Strikovic, E., Urman, A., Hannák, A., de Vreese, C. H., & Esser, F. (2022) · Annals of the International Communication Association · doi:10.1080/23808985.2022.2142149 · Cited by 64 · see citing works
Zarouali, B., Boerman, S. C., Voorveld, H. A. M., & van Noort, G. (2022) · Internet Research · doi:10.1108/intr-01-2021-0049 · Cited by 49 · see citing works
“In the long term, this could contribute to increased polarization in society (Cho et al., 2020; Flaxman et al., 2016; Pariser, 2011).”
BackgroundSeven HCI Grand Challenges Revisited: Five-Year Progress
Stephanidis, C., Salvendy, G., Antona, M., Duffy, V. G., Gao, Q., Karwowski, W., Konomi, S., Nah, F. F., Ntoa, S., Rau, P. P., et al. (2025) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2025.2450411 · Cited by 54 · see citing works
“The capability of algorithms in social media, e.g., filtering, hypernudging, and microtargeting, can be exploited for opinion manipulation (Cho et al., 2020; Christiano, 2022; Nguyen & Vu, 2019).”
Feminism, gender identity and polarization in TikTok and Twitter
Peña-Fernández, S., Ureta, A. L., & Morales-i-Gras, J. (2023) · Comunicar · doi:10.3916/c75-2023-04 · Cited by 41 · see citing works
“On Twitter, meanwhile, there is a greater presence of strategies and dynamics of polarized discourse and segregation, perhaps as a result of the use of algorithms which give rise to the formation of echo chambers and closed communities (Cho et al., 2020).”
Jones-Jang, S. M., & Chung, M. (2024) · New Media & Society · doi:10.1177/14614448221099591 · Cited by 43 · see citing works
“For example, a recent experiment found that YouTube algorithms recommend videos that reinforce viewers’ existing political opinion (Cho et al., 2020).”
BackgroundDemocratization in the age of artificial intelligence: introduction to the special issue
Cupać, J., Schopmans, H., & Ebetürk, I. (2024) · Democratization · doi:10.1080/13510347.2024.2338852 · Cited by 39 · see citing works
Kaluža, J. (2021) · Javnost - The Public · doi:10.1080/13183222.2021.2003052 · Cited by 28 · see citing works
“For example, Cho et al. (2020), who took into consideration the user activity expressed in speci fi c search terms (such as “ Crooked Hillary ” or “ Trump ’ s sexual assault ” ), observed a signi fi cant effect of bubble formation: “ due to the programmed responsiveness to past user interests and…”
BackgroundYoung, D. G., & Miller, J. M. (2023) · Oxford University Press eBooks · doi:10.1093/oxfordhb/9780197541302.013.15 · Cited by 39 · see citing works
Reranking partisan animosity in algorithmic social media feeds alters affective polarization
Piccardi, T., Saveski, M., Jia, C., Hancock, J., Tsai, J. L., & Bernstein, M. S. (2025) · Science · doi:10.1126/science.adu5584 · Cited by 34 · see citing works
“Some studies support this hypothesis, finding that online interactions exacerbate polarization ( 27 ), potentially because of the increased visibility of hostile political discussions ( 28 ), divisive language ( 29 – 33 ), and content that reinforces existing beliefs ( 34 ).”
Kantian Ethics and the Attention Economy
Aylsworth, T., & Castro, C. (2024) · Springer eBooks · doi:10.1007/978-3-031-45638-1 · Cited by 30 · see citing works
Wang, X., & Shang, Q. (2024) · Acta Psychologica · doi:10.1016/j.actpsy.2024.104369 · Cited by 48 · see citing works
Public knowledge of alternative media in times of algorithmically personalized news
Klawier, T., Prochazka, F., & Schweiger, W. (2021) · New Media & Society · doi:10.1177/14614448211021071 · Cited by 29 · see citing works
“Mechanisms operating on these platforms are likely to underlie the same basic logic of pushing attention-grabbing messages, favoring polarizing content (Cho et al., 2020; Schweiger et al., 2019).”
BackgroundLudwig, K., Grote, A., Iana, A., Alam, M., Paulheim, H., Sack, H., Weinhardt, C., & Müller, P. (2023) · Social Science Computer Review · doi:10.1177/08944393221149290 · Cited by 32 · see citing works
“On the other hand, Cho et al. (2020) found YouTube ’ s algorithmic recommendations of political videos, based on the participants ’ search preferences, to affectively polarize users.”
BackgroundLiu, N., Hu, X. E., Savas, Y., Baum, M., Berinsky, A. J., Chaney, A. J. B., Lucas, C., Mariman, R., Benedictis-Kessner, J. D., Guess, A. M., et al. (2025) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2318127122 · Cited by 23 · see citing works
YouTube’s recommendation algorithm is left-leaning in the United States
Ibrahim, H., AlDahoul, N., Lee, S., Rahwan, T., & Zaki, Y. (2023) · PNAS Nexus · doi:10.1093/pnasnexus/pgad264 · Cited by 21 · see citing works
“Nonetheless, there have been several influential studies which aimed to classify YouTube videos into distinct categories (22–25), as well as understand potential radicalization pathways influenced by the recommendation algorithm (21, 26–28).”
BackgroundKilling the golden goose? A framework for regulating disruptive technologies
Bennett, W. L. (2023) · Information, Communication & Society · doi:10.1080/1369118x.2021.1994625 · Cited by 21 · see citing works
“In the political arena, the same proprietary algorithms that make companies so pro fi table also promote user engagement with the polarizing and in fl ammatory content that advocates for democracy, civility and human rights want to regulate (Cho et al., 2020; Tufekci, 2018).”
BackgroundAhmed, S., Madrid‐Morales, D., & Tully, M. (2022) · Online Information Review · doi:10.1108/oir-11-2021-0634 · Cited by 18 · see citing works
Self-citationAhmed, S., & Gil‐López, T. (2022) · Journalism & Mass Communication Quarterly · doi:10.1177/10776990221110113 · Cited by 19 · see citing works
“Cho et al.’s (2020) experimental results lend credence to the premise that heightened selectivity, as a result of the user–algorithm interaction, and user opinion mutually reinforce each other through YouTube’s personalized algorithmic recommendations.”
InfluentialSelf-citationDoes algorithmic filtering lead to filter bubbles in online tourist information searches?
Gong, Y., Schroeder, A., Pan, B., Sundar, S. S., & Mowen, A. J. (2023) · Information Technology & Tourism · doi:10.1007/s40558-023-00279-4 · Cited by 17 · see citing works
“This finding supports our scenario, as previous empirical studies have found that personalization algorithms can reinforce polarized opinions (Cho et al. 2020; Rhodes 2022).”
Calderon, E. D. V., James, T. L., & Lowry, P. B. (2023) · Data and Information Management · doi:10.1016/j.dim.2023.100042 · Cited by 16 · see citing works
“Much has been made of Facebook’s newsfeed algorithm’s potential to influence the democratic process, but everyday phenomena, such as how much one pays for products on Amazon or for a ride on Uber (i.e., dynamic pricing), are also subject to algorithmic decisions [22, 31, 32, 38, 121].”
BackgroundMethodZafrilla, P. J. P. (2021) · Recerca Revista de pensament i anàlisi · doi:10.6035/recerca.4661 · Cited by 16 · see citing works
Huang, B. (2024) · Current Psychology · doi:10.1007/s12144-024-06106-1 · Cited by 15 · see citing works
“The impact of this divergence in access to information is profound (Cho et al., 2020 ).”
The effect of media on opinion formation
Lee, W., Yang, S.-G., & Kim, B. J. (2022) · Physica A Statistical Mechanics and its Applications · doi:10.1016/j.physa.2022.127075 · Cited by 15 · see citing works
Kelm, O., Neumann, T., Behrendt, M., Brenneis, M., Gerl, K., Marschall, S., Meißner, F., Stefan Harmeling, B., Vowe, G., & Ziegele, M. (2023) · Computers in Human Behavior Reports · doi:10.1016/j.chbr.2023.100343 · Cited by 25 · see citing works
Protected how? Problem representations of risk in the General Data Protection Regulation (GDPR)
Padden, M., & Öjehag‐Pettersson, A. (2021) · Critical Policy Studies · doi:10.1080/19460171.2021.1927776 · Cited by 14 · see citing works
“…worthiness of ‘public interest’ is likely to become less so in societies characterized by increasing polarization, which has been shown, ironically, to be heightened by ‘recommender algorithms’ used by streaming services such as YouTube and which are based on user profiles (Cho et al. 2020).”
BackgroundMethodYu, H., Ryoo, Y., & Han, E. (2023) · Internet Research · doi:10.1108/intr-09-2021-0677 · Cited by 10 · see citing works
Prefrontal suppression in short-video viewing: unraveling the neural correlates of self-controlPreprint
Su, C., Teng, B., Zhou, H., Geng, F., & Hu, Y. (2023) · medRxiv · doi:10.1101/2023.10.30.23296738 · Cited by 14 · see citing works
“If not properly monitored, algorithmic recommendations tailored based on user’s behavior might have the potential to influence democracy, privacy and mental health 73,74 .”
BackgroundPolarization in Social Media: A Virtual Worlds-Based Approach
Jacob, D., & Banisch, S. (2023) · Journal of Artificial Societies and Social Simulation · doi:10.18564/jasss.5170 · Cited by 12 · see citing works
Affective Political Polarization and Hate Speech: Made for Each Other?
Стукал, Д., Ахременко, А. С., & Petrov, A. (2022) · RUDN Journal of Political Science · doi:10.22363/2313-1438-2022-24-3-480-498 · Cited by 11 · see citing works
Hu, A., & Ou, M. (2025) · Telematics and Informatics · doi:10.1016/j.tele.2025.102291 · Cited by 18 · see citing works
Coppolillo, E., Mungari, S., Ritacco, E., Fabbri, F., Minici, M., Bonchi, F., & Manco, G. (2025) · Information Processing & Management · doi:10.1016/j.ipm.2025.104125 · Cited by 11 · see citing works
Epstein, R., & Flores, A. (2024) · PLoS ONE · doi:10.1371/journal.pone.0303036 · Cited by 15 · see citing works
Ethical ambiguity and complexity: tech workers’ perceptions of big data ethics in China and the US
Di, D. (2023) · Information Communication & Society · doi:10.1080/1369118x.2023.2166357 · Cited by 10 · see citing works
“Perhaps most worrying is that some tech companies are cavalier about data privacy (Fu, 2019), particularly those that depend for their revenues on signi fi cant social discord and polarization (Cho et al., 2020; Wang, 2015).”
BackgroundMattis, N. M., Masur, P. K., Möller, J. E., & van Atteveldt, W. (2021) · doi:10.31235/osf.io/wvxf5 · Cited by 9 · see citing works
Public Service Media in the Age of Platformization of Culture and Society
Baldini, T. B., López, J. M. T., & Barrientos–Báez, A. (2021) · Palgrave global media policy and business · doi:10.1007/978-3-030-56466-7_3 · Cited by 10 · see citing works
Zhu, Z., Zhang, N., Ding, M., & Chen, L. (2023) · Information Systems Journal · doi:10.1111/isj.12486 · Cited by 11 · see citing works
“Algorithms can thus reduce political selectivity if the range of search terms goes beyond users' political beliefs and identity (Cho et al., 2020; Rekker, 2021).”
BackgroundLee, S., & Kim, K. (2023) · Mass Communication & Society · doi:10.1080/15205436.2023.2272846 · Cited by 10 · see citing works
“As news media systems and journalism practices have changed considerably (e.g., increased user selectivity, algorithmic recommendations, and proliferation of partisan media), it is easier for citizens to encounter partisan news (Bennett & Iyengar, 2008; Cho et al., 2020; Stroud, 2011).”
Sun, M., & Meng, S. (2023) · Youth & Society · doi:10.1177/0044118x231162875 · Cited by 10 · see citing works
“There are concerns, especially in the west, about algorithm-induced ideo-logical polarization through the self-reinforcement of users’ opinions or the dissemination of extremist propaganda (Cho et al., 2020).”
BackgroundHe, J., Gao, B., & Wang, Y.-T. (2024) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2024.2423334 · Cited by 10 · see citing works
“Through precise algorithmic recommendations, users’ satisfaction with media content experiences is heightened, which in turn strengthens the PSI relationship between users and media figures (Cho et al., 2020).”
The Authority to Moderate: Social Media Moderation and its Limits
Kashyap, B., & Formosa, P. (2023) · Philosophy & Technology · doi:10.1007/s13347-023-00685-w · Cited by 8 · see citing works
“…undermine and disrespect user autonomy (Sahebi & Formosa, 2022), lead to widespread and serious mental health issues in vulnerable populations, especially young people (Klein, 2023), and promote the circulation of misinformation (Barnes, 2022) and magnify affective polarisation (Cho et al., 2020).”
BackgroundLudwig, K., Müller, P., Nikolajevic, N., & Grote, A. (2025) · Information Communication & Society · doi:10.1080/1369118x.2024.2435998 · Cited by 8 · see citing works
“The relatively small effect sizes of the present experiment are in line with other studies using real NRS as stimuli (Cho et al., 2020; Kelm et al., 2023) and may not be directly comparable to those of classical experiments.”
Allen, L. K., DeLiema, D., & Kendeou, P. (2025) · Educational Psychologist · doi:10.1080/00461520.2025.2588213 · Cited by 7 · see citing works
“Algorithms reinforce existing opinions by interacting with users’ search terms (Cho et al., 2020), shaping selective news exposure along political lines (Flamino et al., 2023; Guess, 2021), and amplifying ideologically specific social media spaces (Corsi, 2024; Quelle & Bovet, 2025).”
Socializing the political: rethinking filter bubbles and social media with Hannah Arendt
Daus, Z. (2024) · Ethics and Information Technology · doi:10.1007/s10676-024-09759-5 · Cited by 8 · see citing works
People see what they want to see: an EEG study
Yu, H., & Han, E. (2023) · Cognitive Neurodynamics · doi:10.1007/s11571-023-09982-8 · Cited by 6 · see citing works
“…amount of information available online, search engines have become important drivers of how people navigate online, and recommendations based on big data and algorithms are increasingly key to how most users explore, select, and consume information (Beam 2014; Cho et al. 2020; Ricci et al. 2011).”
Mikhaylova, O. R. (2022) · New Media & Society · doi:10.1177/14614448221099172 · Cited by 6 · see citing works
“Because not all vloggers can comply with YouTube’s standards, inequalities are reproduced (Bishop, 2018; Cho et al., 2020).”
BackgroundRomaní, C. C., & Rivera-Vargas, P. (2022) · doi:10.31235/osf.io/yanzw · Cited by 6 · see citing works
Infodemics on Youtube: Reliability of Content and Echo Chambers on COVID-19 (Preprint)Preprint
Marco, N. D., Cinelli, M., & Quattrociocchi, W. (2021) · doi:10.2196/preprints.32258 · Cited by 6 · see citing works
“, loops of extreme/questionable contents reinforced by the algorithm and by individual choices in a sort of vicious circle, is still strongly disputed [35, 36, 37, 38, 39, 40].”
BackgroundBassi, D., Maggini, M. J., Vieira, R. L. R., & Pereira-Fariña, M. (2024) · International Conference on Social Networks Analysis, Management and Security (SNAMS) · doi:10.1109/snams64316.2024.10883781 · Cited by 5 · see citing works
“Furthermore, this polarization could be reinforced by algorithmic recommendation systems that may inadvertently amplify extreme viewpoints [26].”
Lee, S., & Cho, J. (2023) · International Journal of Public Opinion Research · doi:10.1093/ijpor/edad020 · Cited by 5 · see citing works
Self-citationLiao, M. (2023) · Journal of Broadcasting & Electronic Media · doi:10.1080/08838151.2023.2206662 · Cited by 6 · see citing works
“These so-called algorithmic gatekeepers decide the visibility and salience of news content primarily based on user preferences (Cho et al., 2020).”
BackgroundResultInfluentialChernilo, D. (2022) · European Journal of Social Theory · doi:10.1177/13684310221135500 · Cited by 5 · see citing works
“But if they fall short on either technical or ethical grounds, then democratic legitimacy itself suffers reputational costs, which may then be construed as yet another source of its chronic flaws (Cho et al., 2020).”
BackgroundZafrilla, P. J. P. (2022) · Opinião Pública · doi:10.1590/1807-0191202228133 · Cited by 5 · see citing works
Andaluz, J. S., & Rodríguez-Gordo, C. (2021) · Cuadernos salmantinos de filosofía · doi:10.36576/summa.144497 · Cited by 5 · see citing works
“Funny Weapons”: The Norms of Humour in the Construction of Far‐Right Political Polarisation
Toscano, G. B., & Villanueva, C. F. (2025) · Social Inclusion · doi:10.17645/si.10211 · Cited by 4 · see citing works
Análisis de la campaña de desinformación en las Elecciones Brasileñas de 2022
del Vas, R. S., Ruiz-Incertis, R., Acosta, D. G., et al. (2025) · Doxa Comunicación Revista interdisciplinar de estudios de comunicación y ciencias sociales · doi:10.31921/doxacom.n41a2320 · Cited by 4 · see citing works
Segerberg, A., & Magnani, M. (2025) · PLoS ONE · doi:10.1371/journal.pone.0318338 · Cited by 4 · see citing works
“They respond to signals of salience and urgency in engagement metrics and trend tickers [45–47].”
Social Media and Political Polarization: A Panel Study of 36 Countries from 2014 to 2020
Lu, J., Sun, M., & Liu, Z. (2024) · Social Indicators Research · doi:10.1007/s11205-024-03367-y · Cited by 7 · see citing works
“Meanwhile, affective polarization was induced by algorithm recommendation (Cho et al., 2020) and linkage with the accounts of political opponents (Suhay et al., 2018; Allcott et al., 2020).”
Masood, M., Skoric, M. M., & Ahmed, S. (2023) · Journal of Broadcasting & Electronic Media · doi:10.1080/08838151.2023.2298274 · Cited by 4 · see citing works
“Discourse supporting the construction of the temple and, in general, the rights of religious minorities were notably visible online (Ahmed, 2020; Tanzeem, 2020).”
BackgroundSelf-citationPacchioni-Hurtado, V., & Antezana, M. (2023) · Cuadernos del Centro de Estudios de Diseño y Comunicación · doi:10.18682/cdc.vi194.9620 · Cited by 4 · see citing works
Epstein, R., & Flores, A. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4527207 · Cited by 3 · see citing works
Lu, Q., Chen, T., Duan, H., Zhou, X., & Lin, Z. T. (2026) · Journal of Travel Research · doi:10.1177/00472875261441570 · Cited by 4 · see citing works
Soft Power Is Bullshit! How Shared Interests, Not Values, Shape International Relations
Manor, I., & Golan, G. J. (2026) · American Behavioral Scientist · doi:10.1177/00027642251407781 · Cited by 3 · see citing works
Liu, Z.-G. Y. W. J. J.-C. (2025) · Education and Information Technologies · doi:10.1007/s10639-025-13760-w · Cited by 5 · see citing works
“Algorithms from social media and news websites shape the content you will view by assuming what you like based on your previous preferences and interactions; therefore, you are more likely to take in information that supports your previous viewpoint (Cho et al., 2020).”
van der Breggen, M. M., Gonçalves, J., & Boeren, D. (2025) · Journal of Information Technology & Politics · doi:10.1080/19331681.2025.2544653 · Cited by 4 · see citing works
Jiang, T., Sun, Z., & Fu, S. (2025) · Journal of the Association for Information Science and Technology · doi:10.1002/asi.24988 · Cited by 7 · see citing works
“Therefore, the concept of filter bubbles reflects the potential negative consequences of human – algorithm interactions, rather than being solely a biased algorithmic outcome (Areeb et al., 2023; Cho et al., 2020).”
Heiler, P. (2024) · Journal of Econometrics · doi:10.1016/j.jeconom.2024.105856 · Cited by 3 · see citing works
Huang-Isherwood, K. M., Cho, J., Hong, J.-W., & Lee, E. (2024) · Social Science Computer Review · doi:10.1177/08944393241252027 · Cited by 3 · see citing works
“Moreover, fi lter bubbles refer to the vicious cycle of personalized algorithms, including the role of conversational agents, in keeping one bound to past groups and content, and further recommending similar domains without departing from them (Cho et al., 2020; Ojeda, 2021).”
Self-citationCuéllar Rivero, R. (2024) · Revista Española de Ciencia Política · doi:10.21308/recp.64.07 · Cited by 3 · see citing works
Lee, S. S., & Kang, S.-W. (2024) · Korean Journal of Journalism & Communication Studies · doi:10.20879/kjjcs.2024.68.1.010 · Cited by 3 · see citing works
Lee, S. Y., Lee, H., & Park, S. (2024) · International Journal of Public Opinion Research · doi:10.1093/ijpor/edae050 · Cited by 3 · see citing works
Zanetti, L. A., Luvizotto, C. K., Nascimento, F. A., et al. (2023) · doi:10.11648/j.ss.20231204.11 · Cited by 3 · see citing works
Virality, Algorithms, and Illiberal Attacks on the Press: Legitimation Strategies for a New World
Jacobs, R. N. (2020) · Scholars Archive - University at Albany (University at Albany, State University of New York) · doi:10.6092/issn.1971-8853/11190 · Cited by 3 · see citing works
“The user is likely to click on these algorithmically provided links, because they confirmpreexisting biases and interests; when they do this, the algorithm feeds them even more of this kind of content, resulting in ever more narrow and selective exposure (Cho et al., 2020).”
BackgroundClickbait Culture in Internet Journalism: The Digital Way of “Selling News”
Ercan, E. E. (2026) · SELÇUK ÜNİVERSİTESİ İLETİŞİM FAKÜLTESİ AKADEMİK DERGİSİ · doi:10.18094/josc.1697014 · Cited by 3 · see citing works
You, J., Hong, S.-K., Liu, L., Lee, K., & Kwon, H.-Y. (2026) · ACM International Conference on Web Search and Data Mining (WSDM) · doi:10.1145/3773966.3777984 · Cited by 3 · see citing works
Impact of YouTube User‐Generated Content on News Dissemination and Youth Information Reception
Chunqiong, W., Shan, J., Jianhong, S., & Yingqi, L. (2025) · Health Expectations · doi:10.1111/hex.70408 · Cited by 2 · see citing works
The power of cross-cultural communication: reception and impact of Chinese variety shows on YouTube
Zhang, Y. (2025) · Frontiers in Communication · doi:10.3389/fcomm.2025.1542139 · Cited by 2 · see citing works
Strikovic, E., Blassnig, S., Mitova, E., Urman, A., Esser, F., & de Vreese, C. H. (2024) · New Media & Society · doi:10.1177/14614448241263765 · Cited by 6 · see citing works
“However, until now, algorithmic literacy in the context of NRS has not been the focus of research (see Cho et al., 2020; Gran et al., 2021 for exceptions).”
Misinformation Processing Model: How Users Process Misinformation When Using Recommender Algorithms
Shin, D. (2024) · Springer eBooks · doi:10.1007/978-3-031-52569-8_5 · Cited by 2 · see citing works
Misinformation and Generative AI: How Users Construe Their Sense of Diagnostic Misinformation
Shin, D. (2024) · Springer eBooks · doi:10.1007/978-3-031-52569-8_9 · Cited by 3 · see citing works
Zumofen, G. (2023) · Journal of Information Technology & Politics · doi:10.1080/19331681.2023.2221681 · Cited by 2 · see citing works
“Similarly, Cho, Ahmed, Hilbert, Liu, and Luu (2020) experimentally manipulated a real-world search engine.”
InfluentialGriesser, S., Ricklin, A., Kälin, R., & Lu, G. (2023) · IEEE Swiss Conference on Data Science (SDS) · doi:10.1109/sds57534.2023.00026 · Cited by 2 · see citing works
“In fact, more polarising information is viewed more frequently [3] and receives more comments [4].”
BackgroundA K-means Optimized Self-Organizing Map Neural Network Video Recommendation Framework
Liming, Y., & Tianyao, L. (2023) · doi:10.1109/icetci57876.2023.10176756 · Cited by 2 · see citing works
“Therefore, users need to spend a lot of time and effort to search for information, which results in unsatisfactory service [1]–[3].”
BackgroundArandjelović, O. (2023) · Social Sciences · doi:10.3390/socsci12030173 · Cited by 1 · see citing works
Rentsch, S. (2023) · Sozialwissenschaften und Berufspraxis · doi:10.1007/978-3-658-38268-1_2 · Cited by 2 · see citing works
Son, W.-C. (2022) · Philosophy & Technology · doi:10.1007/s13347-022-00573-9 · Cited by 2 · see citing works
“Viewed in this light, algorithms, while helping users navigate a massive amount of information, add to politi‑ cal selectivity that separates people from contradictory views (Cho et al., 2020: 165‑166).”
BackgroundMethodInfluentialJohnson, K. A. (2022) · Communication and Democracy · doi:10.1080/27671127.2022.2141740 · Cited by 2 · see citing works
Processos de Escolha Política, Mediação Anti-democrática e Ambientes Digitais
Conci, L. G. A., & Lopes, J. G. M. (2021) · Direito Público · doi:10.11117/rdp.v18i99.5743 · Cited by 2 · see citing works
Young and Miller, Political Communication in Oxford Handbook of Poli Psych 3rd edPreprint
Young, D. G. (2021) · doi:10.31219/osf.io/mwdtu · Cited by 1 · see citing works
- Authors
- Cho, J., Ahmed, S., Hilbert, M., Liu, B., & Luu, J.
- Affiliations
- University of California, Davis (US)
- Nanyang Technological University (SG)
- Software Engineer, Sacramento, CA, USA
- Software Engineer, San Diego, CA, USA
- Corresponding author
- Cho, Jaeho
- Venue
- Journal of Broadcasting & Electronic Media
- Publisher
- Informa UK Limited
- Published
- 2020-04-27
- Volume
- 64
- Issue
- 2
- Pages
- 150-172
- ISSN
- 0883-8151 (ISSN-L)
- 1550-6878
- OpenAlex
- W3018249611
- Cited by
- 164
- References
- 54
- Citations by year
- 2026: 21 · 2025: 33 · 2024: 33 · 2023: 38 · 2022: 20 · 2021: 17 · 2020: 2
- FWCI
- 23.10
- Type
- journal article
- Language
- en
- Open access
- green
- escholarship.org
- Locations
- Journal of Broadcasting & Electronic Media (publishedVersion)
- eScholarship (California Digital Library) (submittedVersion)
- Topics
- Social Sciences › Communication › Social Media and Politics
- Social Sciences › Sociology and Political Science › Media Influence and Politics
- Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
- Keywords
- algorithmic recommendation, political polarization, YouTube, affective polarization, algorithmic personalization, democracy
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 2026-10-11.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-11.
Cited sources are checked by gpt-6-sol.
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
