Presenting Diverse Political Opinions:How and How MuchSean Munson, Paul ResnickSchool of Information, University of Michigan
Is a polarized world inevitable?
Risks  of  polarization   &   bene1its  of  diversity  
Competing theoriesCHALLENGE AVERSION. People prefermaterial that supports their opinions and avoidchallenging information....
Study goalsINDIVIDUAL PREFERENCES for opinion diversity.EVALUATE SIMPLE PRESENTATION TECHNIQUES for making challenging it...
Study design
Study design: Overall1.  Show people with known political biases a list of links with a predicted percent of agreeable ...
Study design: SubjectsRecruited via Mechanical Turk SUBJECTSRestricted to people in US. ...
Study design: SubjectsRecruited via Mechanical TurkRestricted to people in US. ...
Study design: ArticlesLink source•  Items linked from 500 political blogs; blogs coded as liberal, independent, conserva...
Study design: Experiment2x3 factorial design.•  LIST LENGTH: 8 or 16 item lists•  PRESENTATION Baseline: article tit...
QuestionsASSIGNED TO EITHER: Satisfaction “Suppose this was the front page of a political opinion aggregator. How...
Results
Results: Diversity preferences
Coded as diversity-seeking if…WANT MORE CHALLENGE “e articles in this list showed some of both sides on some issues, but...
Diversity preferences
Model for satisfaction
Support-seeking or challenge-averse?
Support-seeking or challenge-averse?
Support-seeking or challenge-averse? 30 SUBJECTS 10 SUBJECTS
Results: PresentationBASELINE ★ HIGHLIGHT ...
Highlighting
Highlighting + Agreeable First
Results: Presentation
Conclusions & Future Work
Good news and badDIFFERING INDIVIDUAL PREFERENCES for opinion diversity.Challenge aversion is not an inherent human charac...
Future work: PreferencesDistribution of preferencesDo individual preferences change over time?Do individual preferences di...
Future work: PresentationContinue to explore ways to make diversityappealing to challenge-averse people. Field trials for...
Future work: Murky middleInclude people who are less extreme or who don’t fallinto neat categories. ...
thanks!Sean Munson samunson@umich.eduPaul Resnick presnick@umich.edubalance.projects.si.umich.eduis project was supported...
Presenting Diverse Political Opinions: How and How Much (CHI 2010)
Presenting Diverse Political Opinions: How and How Much (CHI 2010)
Presenting Diverse Political Opinions: How and How Much (CHI 2010)
of 33

Presenting Diverse Political Opinions: How and How Much (CHI 2010)

Is a polarized society inevitable, where people choose to be exposed to only political news and commentary that reinforces their existing viewpoints? We examine the relationship between the numbers of supporting and challenging items in a collection of political opinion items and readers’ satisfaction, and then evaluate whether simple presentation techniques such as highlighting agreeable items or showing them first can increase satisfaction when fewer agreeable items are present. We find individual differences: some people are diversity-seeking while others are challenge-averse. For challenge-averse readers, highlighting appears to make satisfaction with sets of mostly agreeable items more extreme, but does not increase satisfaction overall, and sorting agreeable content first appears to decrease satisfaction rather than increasing it. These findings have important implications for builders of websites that aggregate content reflecting different positions.
Published on: Mar 4, 2016
Published in: Technology      News & Politics      
Source: www.slideshare.net


Transcripts - Presenting Diverse Political Opinions: How and How Much (CHI 2010)

  • 1. Presenting Diverse Political Opinions:How and How MuchSean Munson, Paul ResnickSchool of Information, University of Michigan
  • 2. Is a polarized world inevitable?
  • 3. Risks  of  polarization   &   bene1its  of  diversity
  • 4. Competing theoriesCHALLENGE AVERSION. People prefermaterial that supports their opinions and avoidchallenging information. (Selective exposure &homophily)DIVERSITY SEEKING. People preferinformation that contains both challenging andsupporting opinions. (Stromer-Galley 2003, PewIALP 2004)SUPPORT SEEKING. People prefer supportingmaterial but are indifferent to challengingmaterial (Garrett 2009)
  • 5. Study goalsINDIVIDUAL PREFERENCES for opinion diversity.EVALUATE SIMPLE PRESENTATION TECHNIQUES for making challenging items more palatable.
  • 6. Study design
  • 7. Study design: Overall1.  Show people with known political biases a list of links with a predicted percent of agreeable items.2.  Ask them how satisfied they are with the opinions in the set of links
  • 8. Study design: SubjectsRecruited via Mechanical Turk SUBJECTSRestricted to people in US. Age: 34.3 years median: 31 years, standard dev: 11.8 yearsQualification task Gender: 83 men, 87 women•  Demographic questions (gender, zip code, age) Location: 37/50 states•  Political affiliation (7-point scales for party and liberal-conservative)•  3 questions about political knowledgeQuality controlDuring study, subjects randomly re-asked ndemographic questions. Responses from 5 Manipulation check: 30subjects discarded for impossible or improbable Subjects: 40 (satisfaction)replies. 38 (bias)
  • 9. Study design: SubjectsRecruited via Mechanical TurkRestricted to people in US. strongQualification task democrat•  Demographic questions (gender, zip code, age) strong strong•  Political affiliation (7-point scales for party liberal conservative and liberal-conservative)•  3 questions about political knowledge strong republicanQuality controlDuring study, subjects randomly re-askeddemographic questions. Responses from 5subjects discarded for impossible or improbablereplies.
  • 10. Study design: ArticlesLink source•  Items linked from 500 political blogs; blogs coded as liberal, independent, conservative•  Daily: select 40 most-linked stories from the previous 24 hours that had ≥2:1 liberal:conservative or conservative:liberal link ratio•  Filter out tweets, YouTube videos, Wikipedia articles, items not matching predicted bias. Average of 23 items / bias / day remain.Manipulation check•  30 turkers. Shown links, asked if they agree or disagree. Each link seen by ≥3 turkers.•  Kept links to which they reacted consistent with expectation. Discarded others.
  • 11. Study design: Experiment2x3 factorial design.•  LIST LENGTH: 8 or 16 item lists•  PRESENTATION Baseline: article title (linked) + abstract Highlight: Agreeable items highlighted Highlight+Order: Agreeable items highlighted and placed first•  Varying PERCENT OF AGREEABLE ITEMS
  • 12. QuestionsASSIGNED TO EITHER: Satisfaction “Suppose this was the front page of a political opinion aggregator. How would you feel about the viewpoints represented in it?” (5 point Likert scale, Very dissatisfied to very satisfied) Bias “What, if any, is the political bias of this collection?” (5 point Likert scale, Very Liberal to Very Conservative)EVERYONE: Why they gave the rating they did (open-ended) Random demographics check
  • 13. Results
  • 14. Results: Diversity preferences
  • 15. Coded as diversity-seeking if…WANT MORE CHALLENGE “e articles in this list showed some of both sides on some issues, but on other issues like health care was rather one sided. If that and a few other articles had been given two sides I would be completely satisfied. I like to read both sides even though I am mostly conservative.”WOULDN’T WANT LESS CHALLENGE “ere is an even distribution of right and le wing articles. I think it is best to cover both sides of the issue.” “I like that there are views from both Democrats and Republicans and seems to be a great mix of both sides of the fence.” Cohen’s  kappa:  0.89
  • 16. Diversity preferences
  • 17. Model for satisfaction
  • 18. Support-seeking or challenge-averse?
  • 19. Support-seeking or challenge-averse?
  • 20. Support-seeking or challenge-averse? 30 SUBJECTS 10 SUBJECTS
  • 21. Results: PresentationBASELINE ★ HIGHLIGHT ★ HIGHLIGHT + ORDER Ar#cle  #tle   Ar#cle  #tle   Ar#cle  #tle   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.   Ar#cle  #tle   Ar#cle  #tle   Ar#cle  #tle   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.   Ar#cle  #tle   Ar#cle  #tle   Ar#cle  #tle   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.   Ar#cle  #tle   Ar#cle  #tle   Ar#cle  #tle   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   Story  abstract  would  go  here.  And  so   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.   on.  It  would  say  stuff.  And  have  opinion.
  • 22. Highlighting
  • 23. Highlighting + Agreeable First
  • 24. Results: Presentation
  • 25. Conclusions & Future Work
  • 26. Good news and badDIFFERING INDIVIDUAL PREFERENCES for opinion diversity.Challenge aversion is not an inherent human characteristic, butneither is diversity seeking.SIMPLE PRESENTATION TECHNIQUES, such as highlighting andordering agreeable items, do not make diverse opinions moreappealing to challenge-averse individuals.
  • 27. Future work: PreferencesDistribution of preferencesDo individual preferences change over time?Do individual preferences differ by topic?
  • 28. Future work: PresentationContinue to explore ways to make diversityappealing to challenge-averse people. Field trials for ordering Social influence techniques Ideas derived from NewsCube (Park et al, CHI 2009): show agreeable item first or prominently, then indicate that there are differing views.
  • 29. Future work: Murky middleInclude people who are less extreme or who don’t fallinto neat categories. strong democrat strong strong liberal conservative strong republican
  • 30. thanks!Sean Munson samunson@umich.eduPaul Resnick presnick@umich.edubalance.projects.si.umich.eduis project was supported by NSF Award IIS-0916099.

Related Documents