Socio-Temporal Dynamics in Peer Interaction Events

Socio-Temporal Dynamics in Peer Interaction Events


Bodong Chen

March 25, 2020


  1. Socio-Temporal Dynamics in Peer Interaction Events Bodong Chen, University of

    Minnesota Oleksandra Poquet, University of South Australia #LAK20, March 25, 2020 [Link to paper:]
  2. Social Participation in online/blended learning Photo Credit

  3. Social Participation Is linked to: Academic performance “Soft skills” Sense

    of belonging Social capital … Photo Credit (Jo, Park & Lee, 2017; Joksimović, Gasević, Kovanovic, Riecke & Hatala, 2015; Zheng & Warschauer, 2015)
  4. Social Participation & Peer Interaction Pedagogical models Digital tools, social

    media Informal spaces … Various attempts to foster Photo Credit (Dennen, 2005; Marbouti & Wise, 2016; Sadowski, Pediaditis & Townsend, 2017; Wang, 2005; Wise & Chiu, 2011)
  5. There is a participation gap in classroom discussions (Chen &

    Huang, 2019; Vaquero & Cebrian, 2013)
  6. Social Participation & Peer Interaction A methodological gap in analyzing

    Photo Credit
  7. A major, enduring question: What are the mechanisms of social

    interaction in asynchronous online discussions? Photo Credit
  8. Framework

  9. Research question: How does socio-temporal dynamics contribute to social interaction

    in asynchronous online discussions? Photo Credit
  10. Study Context Courses n n_posts n_activity 2711 Technology and Ethics

    21 276 1137 2713 Info Systems 82 795 3470 2758 Operations Management 29 110 829 2768 IT and Solutions A 33 277 1031 2839 IT and Solutions B 20 189 652 2853 Innovation Management A 43 159 1379 3446 Innovation Management B 43 121 1389 a pin reactions replies A board
  11. Data Analysis 1) Temporal profiles of learners Logs each day

    of the week Finite mixture models Time-consistent posters Last-minute posters
  12. Sequential Structural Signatures (SSSs) 2) Relational Event Modeling (REM) i

    j i j i j i j t (Butts, 2008; Leenders et al., 2015; Pilny et al., 2016) Past relational events Actor attributes Exogenous factors The next event
  13. Effects Visualization Description Past relational events Nodal: Total Degree (NTDegRec)

    Dyadic: Familiarity (RSndSnd) Reciprocity (RRecSnd) Triadic: Outbounding shared partner (OSPSnd) Well-connected students to receive more interactions Past interactions to happen again Past interactions to be reciprocated Shared outbounding partners → direct interactions Modeled SSSs and Actor Attributes i j i j k i j t t’ i j t t’
  14. Effects Description Past relational events (cont’d) Immediate Effects: Participation Shifts

    (Gibson, 2005): AB → BA AB → BY AB → AY AB → XB Reciprocation Pay-it-forward or hand-off Activity bursts, persistence of source Popularity, persistence of target Actor attributes (controls) Time-consistent Role Being time-consistent (vs. last-minute) Instructor (vs. student) Modeled SSSs and Actor Attributes (cont’d)
  15. Findings

  16. Course 1 How the interaction networks look like?

  17. Relational Event Models

  18. Key REM results (1) -- node/actor attributes • Time-consistent learners

    → Positive tendency to reply to others (4/7) → Mixed tendency to receive replies (2+,1- / 7) • Being the instructor → Positive tendency to reply to others (4/7) → Negative tendency to receive replies (6/7)
  19. Key REM results (2) -- socio-temporal structures • Learner total

    degree → Positive tendency to receive replies (7/7) • Past A→B ties → Positive tendency to A→B ties (7/7) → Positive tendency to B→A ties (6/7) • A and B have an outbounding shared partner (e.g., A→C←B) → Only weak positive tendency to A→B ties (4/7)
  20. Key REM results (3) -- immediate effects • One A→B

    tie → Strong positive tendency to have a A→Y tie immediately next (7/7) → Positive tendency to have a Y→B tie immediately next (6/7) → Positive tendency to have a B→A tie immediately next (4/7) → Mixed results about having a B→Y tie immediately next (3+,1- / 7)
  21. Key Takeaways • Peer interaction is shaped by a range

    of factors. • The analysis of peer interaction can benefit from focusing on relational events rather than relational states. • Social dynamics and temporal dynamics in peer interaction networks are intertwined.
  22. Limitations and Future Work • Data sources were limited to

    timestamped activity logs ◦ No information about demographics ◦ No information about post content • Combine relational event modeling w/ discourse analysis • Develop and contrast pedagogical strategies • Novel statistics for learning analytics tools
  23. Thank You! Bodong Chen Email: Twitter: @bod0ng Website:

    Sasha Poquet Email: Twitter: @choux Website: