Skip to content

Latest commit

 

History

History
103 lines (94 loc) · 3 KB

TODO.md

File metadata and controls

103 lines (94 loc) · 3 KB

To do list:

Benevolence study

Initial

  • Finalize design
  • Preregistration
  • Run
  • Data analysis
  • Refine paradigm
  • Preregistration
  • Run
  • Data analysis
  • Redesign for explicit benevolence manipulation
  • Preregistration (not done)
  • Run
  • Data analysis
  • Redesign explicit benevolence manipulation
    • Clearer manipulation
      • At advisor introduction
        • "will sometimes try to mislead"
      • "You are in Group 1" displayed throughout
    • Actually provide occasionally bad advice from outgroup advisor
    • Improve advice slightly (inc. for training advisor)
  • Preregistration (omitted)
  • Run
  • Data analysis
  • Redesign based on clearer manipulation
    • Pilot
    • Data Analysis
    • Preregistred replication
  • Advisor population version
    • Pilot
    • Preregistration
    • Data analysis

Registered Report

  • Identify journal(s)
  • RR stage 1
  • Revise RR
  • Run RR
  • RR stage 2

Confidence mapping study

  • Devise confidence mapping manipulation
  • Design study
  • Pilot
  • Rework
  • Preregistration

Core studies

  • Clarify marker usage (confidence)
  • Redesign core studies
    • One marker width
    • Post-decision confidence judgement
  • Implement Agree-in-confidence/uncertainty advisors
    • Discrete distances from participant marker 3, 5, 7 | 9, 11, 13
    • AiC uses first set where high conf
    • AiU uses second set where high conf
    • Tests on trials where 8 used regardless of confidence (offbrand)
    • Integration tests
  • Run AiC/AiU study

Advisor choice

  • Implement advisor choice
  • Run advisor choice for Agree/Accurate
    • Choice trials should give offbrand advice, so influence can be compared on choice trials (how to compare for participants who only pick one advisor?)
  • Run advisor choice for AiC/AiU

Non-advice-takers

  • Work out how to recruit people who don't take advice
  • Develop paradigm for testing people's base levels of advice-seeking/taking
    • Time-taking alternative to getting advice
      • See advice but opt out of using it before/after seeing it?

Agent-based models

  • Build model of agents' advisor-updating and source selection
  • Run dynamic networks of interacting agents
  • Characterize resulting network structures based on agent properties (e.g. model components)
  • Does including non-advice-takers change network structure?
    • Model as normal agents with particularly low update rates?

Real-life networks

  • Characterize real-life social networks in terms used for ABMs above
  • Compare RL networks to ABM networks

Write thesis

  • Learn Oxforddown
  • Introduction
    • Literature review
    • Core concepts
  • Psychology of advice
    • Introduction
    • Advice-taking
    • Source selection
  • Contextual factors
    • Introduction
    • Models
    • Study
  • Interacting agents
    • Agent-based models
    • Real-world networks in comparison
  • Conclusion
  • References