From IMC wiki
Jump to: navigation, search

Group size

  • In theory can be any size; practical issues mean easier to recruit and test people in 1s 2s or 3s...?
  • Also, how do you compare individuals to groups - lots of different ways in the literature (better than best member etc) but...
  • Do we get participants to contribute in more than one sized group (for same/different tasks)?

Group structure

  • Manipulate who sees information from whom and whether this is reciprocal or not and whether participants are aware of it or not
  • Could be either certain people's contributions are not available to everyone, or just be certain things not shared so that e.g. one person gets an impoverished view of another's contributions?

Meeting: 20/02/2013

Looked at the different possible structures (star, wheel, totally interconnected, line etc).

  • Questions:
    • What is the contrast we are looking at? NB needs to be simple for the purposes of this pilot project
      • Centralised vs decentralised networks? (i.e. star versus totally interconnected)
      • Dynamic vs static networks? (start in one state e.g. circle and randomly mutate a link)
      • Levels of reciprocity?
    • What are the outcome measures
      • At an individual level are confounded by different cognitive loads
      • At a group level depends on task (e.g. Alt Uses is novelty/number of different solutions, not solution time/accuracy)
      • Conversational moves (how they come to solutions in different structures)
  • Practical considerations
    • One window per person/per conversation or in total (depending on the decision this means different amounts of stuff on different people's screens
    • Not really constrained by what the chat tool can do (except ruling out mass participation mechanical turk type implementation - for now!) in terms of who sees what.
    • Need to decide appropriate task (depends on what outcome measures we are interested in)

Boils down to: Can we come up with a robust testable hypothesis for some subset of this stuff that we're actually interested in?!

Communication processes

  • How much communication do groups need to partake in to have an advantage (if they get one, that is)
  • What sort of tasks are facilitated by the interactiveness of dialogue in addition to having more than one person partaking in the task
    • i.e. for groups is it just that there is input from more than one person, or is it also in the negotiation (this probably depends on the task)
    • could either let them say anything they wanted (including feedback etc) or only offer potential solutions.
    • contrast interactive exposure to another persons ideas vs non-interactive - i.e. either talking to someone, or see the ideas generated (in same time as generated) by a previous participant.
  • Does the type/strength of feedback matter ("okay", "uh", "what?") (downgrades/upgrades/added fake CRs/grounding cues)