Televised Debates, Second Screens,and Filter Bubbles

presented at the EPSA Annual Conference 2018

Simon Richter1 & Prof. Dr. Thorsten Faas1

1Otto-Suhr-Institut

23.06.2018

TV Debates in Context: Past & Present

TV Debates in Context: Past & Present

Second Screening

Second screening

= ”bundle of practices that involve integrating, and switchingacross and between, live broadcast and social media”(Vaccari et al. 2015)

I increasingly popular in general

I most prominent during media events

I motivations for second screening: discuss, get furtherinformation and gauge others’ opinions

Filter Bubbles

Filter Bubbles= communicative spaces in which “content is selected byalgorithms according to a viewer’s previous behaviors” (Bakshy et al.

2015) , thereby providing “content an individual is likely to agreewith” (Flaxman et al. 2016).

I Homophily on social media platforms is a thing, ...

I ... but there is a fair chance that users get confronted withattitude-discordant contents.

I Unexplored: Effects of Filter Bubbles on perception of politicalinformation. Why is that?

I Idiosyncratic information environments: unobservable fromoutside and hard to generalize their features

I Endogeneity: self-selection into exposure makes effectestimation through purely observational data pretty muchimpossible

Filter Bubbles

Filter Bubbles= communicative spaces in which “content is selected byalgorithms according to a viewer’s previous behaviors” (Bakshy et al.

2015) , thereby providing “content an individual is likely to agreewith” (Flaxman et al. 2016).

I Homophily on social media platforms is a thing, ...

I ... but there is a fair chance that users get confronted withattitude-discordant contents.

I Unexplored: Effects of Filter Bubbles on perception of politicalinformation. Why is that?

I Idiosyncratic information environments: unobservable fromoutside and hard to generalize their features

I Endogeneity: self-selection into exposure makes effectestimation through purely observational data pretty muchimpossible

Filter Bubbles

Filter Bubbles= communicative spaces in which “content is selected byalgorithms according to a viewer’s previous behaviors” (Bakshy et al.

2015) , thereby providing “content an individual is likely to agreewith” (Flaxman et al. 2016).

I Homophily on social media platforms is a thing, ...

I ... but there is a fair chance that users get confronted withattitude-discordant contents.

I Unexplored: Effects of Filter Bubbles on perception of politicalinformation. Why is that?

I Idiosyncratic information environments: unobservable fromoutside and hard to generalize their features

I Endogeneity: self-selection into exposure makes effectestimation through purely observational data pretty muchimpossible

Filter Bubbles

Filter Bubbles= communicative spaces in which “content is selected byalgorithms according to a viewer’s previous behaviors” (Bakshy et al.

2015) , thereby providing “content an individual is likely to agreewith” (Flaxman et al. 2016).

I Homophily on social media platforms is a thing, ...

I ... but there is a fair chance that users get confronted withattitude-discordant contents.

I Unexplored: Effects of Filter Bubbles on perception of politicalinformation. Why is that?

I Idiosyncratic information environments: unobservable fromoutside and hard to generalize their features

I Endogeneity: self-selection into exposure makes effectestimation through purely observational data pretty muchimpossible

Televised Debates and Filter Bubbles

Televised Debates and Filter Bubbles

Televised Debates and Filter Bubbles

RQ1:Do the subjects “accurately“ perceive the tone of the filter bubblethey are in?

I Identifying the tone of filter bubble = highly complex taskI Different modes of information processing possible (Schulz &

Roessler 2012)

I quasi-statistical senseI looking-glass perception

Televised Debates and Filter Bubbles

RQ1:Do the subjects “accurately“ perceive the tone of the filter bubblethey are in?

I Identifying the tone of filter bubble = highly complex task

I Different modes of information processing possible (Schulz &

Roessler 2012)

I quasi-statistical senseI looking-glass perception

Televised Debates and Filter Bubbles

RQ1:Do the subjects “accurately“ perceive the tone of the filter bubblethey are in?

I Identifying the tone of filter bubble = highly complex taskI Different modes of information processing possible (Schulz &

Roessler 2012)

I quasi-statistical senseI looking-glass perception

Televised Debates and Filter Bubbles

RQ1:Do the subjects “accurately“ perceive the tone of the filter bubblethey are in?

I Identifying the tone of filter bubble = highly complex taskI Different modes of information processing possible (Schulz &

Roessler 2012)

I quasi-statistical sense

I looking-glass perception

Televised Debates and Filter Bubbles

RQ1:Do the subjects “accurately“ perceive the tone of the filter bubblethey are in?

I Identifying the tone of filter bubble = highly complex taskI Different modes of information processing possible (Schulz &

Roessler 2012)

I quasi-statistical senseI looking-glass perception

Televised Debates and Filter Bubbles

Televised Debates and Filter Bubbles

Televised Debates and Filter Bubbles

RQ2:Do the biased information environments influence the perceptionof the candidates’ performances?

I Televised debates are highly complex → need for heuristics

I Pre-existing attitudes towards candidates/partiesI Viewers geared by other users’ opinions (social influence

theory):

I Political attitudes in general (Levitan & Verhulst 2016) andcandidate evaluation in televised debates shown to besusceptible to social influence

I Social influence can occur in computer-mediatedcommunication spaces (see Maruyama et al. 2017)

Televised Debates and Filter Bubbles

RQ2:Do the biased information environments influence the perceptionof the candidates’ performances?

I Televised debates are highly complex → need for heuristics

I Pre-existing attitudes towards candidates/partiesI Viewers geared by other users’ opinions (social influence

theory):

I Political attitudes in general (Levitan & Verhulst 2016) andcandidate evaluation in televised debates shown to besusceptible to social influence

I Social influence can occur in computer-mediatedcommunication spaces (see Maruyama et al. 2017)

Televised Debates and Filter Bubbles

RQ2:Do the biased information environments influence the perceptionof the candidates’ performances?

I Televised debates are highly complex → need for heuristics

I Pre-existing attitudes towards candidates/parties

I Viewers geared by other users’ opinions (social influencetheory):

I Political attitudes in general (Levitan & Verhulst 2016) andcandidate evaluation in televised debates shown to besusceptible to social influence

I Social influence can occur in computer-mediatedcommunication spaces (see Maruyama et al. 2017)

Televised Debates and Filter Bubbles

RQ2:Do the biased information environments influence the perceptionof the candidates’ performances?

I Televised debates are highly complex → need for heuristics

I Pre-existing attitudes towards candidates/partiesI Viewers geared by other users’ opinions (social influence

theory):

I Political attitudes in general (Levitan & Verhulst 2016) andcandidate evaluation in televised debates shown to besusceptible to social influence

I Social influence can occur in computer-mediatedcommunication spaces (see Maruyama et al. 2017)

Study Design

I Methodological innovation: Laboratory Live-Experiment onGerman televised debate 2017

I Between-subjects design with three different twitter wallscontaining real tweets

I Sample: 119 participants highly educated and rather young,balanced in gender

I Random assignment worked, but coincidental deviations inparty ID

Study Design

I Methodological innovation: Laboratory Live-Experiment onGerman televised debate 2017

I Between-subjects design with three different twitter wallscontaining real tweets

I Sample: 119 participants highly educated and rather young,balanced in gender

I Random assignment worked, but coincidental deviations inparty ID

Study Design

I Methodological innovation: Laboratory Live-Experiment onGerman televised debate 2017

I Between-subjects design with three different twitter wallscontaining real tweets

I Sample: 119 participants highly educated and rather young,balanced in gender

I Random assignment worked, but coincidental deviations inparty ID

Study Design

I Methodological innovation: Laboratory Live-Experiment onGerman televised debate 2017

I Between-subjects design with three different twitter wallscontaining real tweets

I Sample: 119 participants highly educated and rather young,balanced in gender

I Random assignment worked, but coincidental deviations inparty ID

Study Design

RQ1: Perception of the Filter Bubble tone

QuestionRecalling the tweets you could observe during the debate:Altogether, how was [Angela Merkel/Martin Schulz] portrayed inthose messages from your point of view?

1 = very negative; 5 = very positive

RQ1: Perception of the Filter Bubble tone

RQ2: Effects on candidate evaluation

QuestionAltogether, how did [Angela Merkel/Martin Schulz] perform duringthe debate?

1 = very bad; 5 = very good

RQ2: Effects on candidate evaluation

Conclusion

I Subjects are able to determine the tone of their filter bubblein a quasi-statistical manner; however: unexplained variationmerits further investigation

I Filter bubble effect only for Martin Schulz → effectscontingent on external factors (e.g. pre-existing knowledgeabout candidate)

I Implications: Filter Bubble effects opening ways to influencepolitical attitudes through organized collective actions onsocial media channels (”Hijacking the filter bubble”)

I Upcoming: Survey Experiment (positive vs. negative tweets)

Conclusion

I Subjects are able to determine the tone of their filter bubblein a quasi-statistical manner; however: unexplained variationmerits further investigation

I Filter bubble effect only for Martin Schulz → effectscontingent on external factors (e.g. pre-existing knowledgeabout candidate)

I Implications: Filter Bubble effects opening ways to influencepolitical attitudes through organized collective actions onsocial media channels (”Hijacking the filter bubble”)

I Upcoming: Survey Experiment (positive vs. negative tweets)

Conclusion

I Subjects are able to determine the tone of their filter bubblein a quasi-statistical manner; however: unexplained variationmerits further investigation

I Filter bubble effect only for Martin Schulz → effectscontingent on external factors (e.g. pre-existing knowledgeabout candidate)

I Implications: Filter Bubble effects opening ways to influencepolitical attitudes through organized collective actions onsocial media channels (”Hijacking the filter bubble”)

I Upcoming: Survey Experiment (positive vs. negative tweets)

Conclusion

I Subjects are able to determine the tone of their filter bubblein a quasi-statistical manner; however: unexplained variationmerits further investigation

I Filter bubble effect only for Martin Schulz → effectscontingent on external factors (e.g. pre-existing knowledgeabout candidate)

I Implications: Filter Bubble effects opening ways to influencepolitical attitudes through organized collective actions onsocial media channels (”Hijacking the filter bubble”)

I Upcoming: Survey Experiment (positive vs. negative tweets)

ReferencesI Bakshy E, Messing S, Adamic LA. 2015. Political science.

Exposure to ideologically diverse news and opinion onFacebook. Science (New York, N.Y.) 348 (6239):1130-32.

I Flaxman S, Goel S, Rao JM. 2016. Filter Bubbles, EchoChambers, and Online News Consumption. PUBOPQ 80(S1):298-320.

I Hahn, Kyu S., Hye-Yon Lee, Seyong Ha, Seulgi Jang, andJoonwhan Lee. 2017. “The Influence of “Social Viewing” onTelevised Debate Viewers’ Political Judgment.” PoliticalCommunication 42 (1): 1-19.

I Levitan LC, Verhulst B. 2016. Conformity in Groups. TheEffects of Others’ Views on Expressed Attitudes and AttitudeChange. Political Behavior 38 (2):277-315.

I Maruyama, Misa. 2017. “Social Watching a Civic Broadcast.”In the 2017 ACM Conference, eds. Charlotte P. Lee, StevePoltrock, Louise Barkhuus, Marcos Borges and WendyKellogg, 794-807.

References

I Schulz A, Roessler P. 2012. The Spiral of Silence and theInternet. Selection of Online Content and the Perception ofthe Public Opinion Climate in Computer-MediatedCommunication Environments. International Journal of PublicOpinion Research 24 (3):346-67.

I Vaccari C, Chadwick A, O’Loughlin B. 2015. Dual Screeningthe Political: Media Events, Social Media, and CitizenEngagement. Journal of Communication 65.