Choice-Based Conjoint (CBC) Workshop
by Peter Kurz and Winfried Steiner
Mercure Hotel Dortmund Centrum
Olpe 2, 44135 Dortmund
|Start:||Tuesday,||October 8, 2019||09:00|
|End:||Friday,||October 11, 2019||16:30|
This four-day workshop introduces participants to the state-of-the-art as well as to advanced topics in choice-based conjoint (CBC) analysis. Days 1 and 2 provide a profound overview of all basic data collection and data analysis steps in order to conduct a CBC study, like questionnaire design, constructing efficient CBC designs, and estimating utilities using the multinominal logit model including latent class analyses. Day3 addresses Hierarchical Bayes models, provides guidelines for market simulations based on estimated conjoint utilities, and discusses pros and cons of related choice rules. Day 4 is planned as a colloquium. It provides participants with the opportunity to present their own (dissertation) research and to discuss possible CBC approaches with the other participants and the instructors to answer the research questions. Alternatively or additionally, case studies are discussed how to set up CBC studies for concrete problem settings.
All contents are illustrated with examples and exercises, and participants will learn how to apply the CBC approach by means of the Sawtooth software.
The workshop will cover all pressing questions concerning the design of a CBC study as well as recent methodical developments for analyzing CBC data, including:
Participants will receive a 30-days fully functional version of the Sawtooth software modules used in the workshop (Lighthouse Studio, Latent class, CBC/HB, ASM).
- how to set critical parameters like the number of attributes per choice set, the number of alternatives per choice set, the number of choice sets per respondent, and the number of respondents,
- how to assess criteria that affect the efficiency of a CBC design (orthogonality, level overlap, level balance, utility balance), whether to choose fixed versus randomized tasks, when to focus on main effects versus interaction effects,
- how to handle specific interaction- and cross-effects with alternative-specific designs (ASD − the real discrete choice models),
- how to estimate segment-specific utilities using the latent class approach versus individual utilities using an Hierarchical Bayes approach, and how to evaluate model performances from a statistical and managerial point of view,
- how and when to use which choice rules for market simulation (first choice, logit choice, randomized first choice, random draws),
- how to build optimal conjoint choice models when "price" is the main focus,