Attention and Bias in Social Information Networks
?Download as PPTX, PDF?
4 likes?974 views
The document discusses how attention and biases impact social information networks. It finds that looking time per tweet is short and memory is poor. Personal contacts increase attention and memory compared to including links, retweets, or heavy tweeting. User names influence perception of authors and their content, with good authors getting higher ratings when their name is visible, while average and bad authors see ratings decrease. Credibility of tweets and authors is modestly correlated with perceptions of truth. Factors like friends, name value, and gender can introduce biases. Overall it suggests social media uses peripheral processing and heuristics rather than careful analysis due to limits on attention, memory, and determining truthfulness online.
Attention and Bias in Social Information Networks
- 1. Attention and bias in social information networksScott counts, microsoft research
- 4. Looking time per tweet is short, memory is poor.
- 5. Looking time per tweet is short, memory is poor.
- 6. Looking time per tweet is short, memory is poor.
- 7. Including links, RTs, heavy tweeting all decrease attention and/or interest.
- 8. Including links, RTs, heavy tweeting all decrease attention and/or interest.
- 9. Including links, RTs, heavy tweeting all decrease attention and/or interest.
- 10. Including links, RTs, heavy tweeting all decrease attention and/or interest.
- 11. Personal contacts increase attention and memory.Counts, S., & Fisher, K. (2011). Taking It All In? Visual Attention in Microblog Consumption. In Proc. ICWSM ¡®11.
- 13. Problem statementHow does a user¡¯s name influence perception of her and her content?
- 16. Results ¨C author ratingsFairly bimodal distributionsDownward shift in ratings when non-anonymous
- 17. Results ¨C rating distributionGood author get higher ratings when non-anon.Bad authors hurt most by namesAverage authors similar to good (KL div = .02) but hurt by name (KL div = .23; p < .001)
- 18. Results ¨C ratings & follower countResults tighten up with names: R2 = .16 -> .21High follower count people get biggest boostMiddle group hurtPal, A., & Counts, S. (2011). What¡¯s In a @Name? How Name Value Biases Judgment of Microblog Authors. In Proc. ICWSM ¡®11.
- 24. Credibility and truth*Name type impacts tweet and author credibilityCorrelations between truth and tweet (r = .39) and author (r = .29) modest* Morris, M., Counts, S., Roseway, A., Hoff, A., & Schwartz, J. (2011). Under review.
- 26. Bringing it togetherMinimal visual processing/attentionPoor memory encoding
- 27. Bringing it togetherMinimal visual processing/attentionPoor memory encodingDifficulty in determining truthfulnessSystematic use of heuristics (biases)FriendsName value
- 28. Bringing it togetherMinimal visual processing/attentionPoor memory encodingDifficulty in determining truthfulnessSystematic use of heuristics (biases)FriendsName value** Peripheral processing route **
- 29. Implications
- 30. ImplicationsEffective reach of social media
- 31. ImplicationsEffective reach of social media Information diffusion
- 32. ImplicationsEffective reach of social media Information diffusion Social contagion: Stickiness* (increased adoption and sustained product use) and memory for content* Aral, S., & Walker, D. (2010). Creating Social Contagion Through Viral Product Design: A Randomized Trial of Peer Influence Networks. Management Science.
- 33. Attention and bias in social information networksScott counts, microsoft research
- 34. low level :: your brain on facebook** Fisher, K., & Counts, S. (2010). Your Brain on Facebook: Neuropsychological Associations with Social Versus Other Media. In Proc. ICWSM ¡®10.
- 35. social information networks :: levels of analysisMath/TheorySocial media analyticsComputer-Mediated CommunicationSocial CognitionPhysiological
- 36. Results ¨C factors for bias: genderMost top authors are gender neutral (e.g., Time, Mashable)Men higher than women when anonymous, but drop more when names shownWomen get slight bump when names shownPal, A., & Counts, S. (2011). What¡¯s In a @Name? How Name Value Biases Judgment of Microblog Authors. In Proc. ICWSM ¡®11.
- 37. social information networks :: levels of analysisMath/TheorySocial media analyticsComputer-Mediated CommunicationSocial CognitionPhysiological
- 38. Problem statementHow does a user¡¯s name influence perception of her and her content?
- 39. Problem statementHow does a user¡¯s name influence perception of her and her content?
Editor's Notes
- alshepmcr