Transcript Dia 0

Workshop III
Mental representations
and discrete choice behaviour
State of the art and avenues for future research
Background
Chair:
Theo Arentze
Presenters:
Benedict Dellaert, Geert Wets
Discussants:
Caspar Chorus, Harmen Oppewal
• To discuss new perspectives emerging from
• recent new developments in the area of models and
data collection to integrate mental representations in
discrete choice models
− Dedicated (online) tools provide scaleability
− Apporaches to integrated model development
Agenda
• Paper presentations
− An interactive computer-based interface to support discovery of
individuals’ mental representations and preferences in decision
problems - Geert Wets
Discussant: Caspar Chorus
− Online measurement of mental representations of complex
decision problems - Benedict Dellaert
Discussant: Harmen Oppewal
• Discussion of several key themes
• Wrap-up and conclusions
Traditionally
Cognition
Choice
Data
collection
New
Attitudes / concepts /
values not related to a
specific choice task
Mental representation of a
specific choice problem
Trade-off between
attributes is invariant
Situation-dependent
motivations / benefits
Qualitative in a separate
stage (e.g., focus groups)
Integrated with choice
tasks
Dedicated (online) tools
provide scaleability
Model developments
Bayesian network model
Joint MR and choice
model
Reveals what is underneath
individuals choice behavior
Discussion
• Why this approach – what are the gains?
• Improved prediction of choice behavior?
− One-layer multi-attribute models may fit equally well?
− Changes in behavior in (radically) new situtations
• Better understanding of the behavior
− Extends the idea of individual-level modeling
− Particularly relevant when interpretations of
consequences are a-priori unclear
−
−
−
−
MRs of risk
MRs of strategic behavior in decision questions
MRs of political program choices
Influence of group discussion / assumed role
− Offers new meaningful ways for segmentation
Discussion cont’d
• Supporting policy making
− ‘More buttons to push’
− Communication of policies – effective framing
− Innovation in policy making
Discussion cont’d
• What is it that we measure? – caveats
• Explanatory value of elicited cognitions?
− More than just rationalisations / justifications / errorprone reconstructions?
• Influence of measurement
− Framing effects and order effects
• Combine with additional data
− Non-verbal approaches (e.g., response time)
− Information search behaviour (or eye tracking)
Conclusions
• Operational models and online data collections tools
are now available to integrate mental representations
and discrete choice models
• Has value particularly in terms of better
understanding, policy making and communication
• Opened up new perspectives for application areas
and further development