Author: Gary Bennett
Best-Worst Case 2 (BWC2) models continue to experience huge growth. These models are a half-way house between Max-Diff models and full Choice Based Conjoints. They enable levels of features of products and services to be directly traded off against each other to determine what drives preference.
Example:
Q: Which of these is most likely to encourage you to book with this airline? And which is least likely to encourage you to book with this airline?
Advantages of Best-Worst Case2 approach:
- Can incorporate more attributes than a conventional CBC, with attributes rotated in and out of exercises
- All option levels are placed on the same scale – which is not the case with conventional choice-based conjoint where the variants for option can only be compared WITHIN each option
- It stops one option from dominating the others (as usually only between 3-6 attributes appear per choice in the example) which can be problematic in a conventional DCM/CBC
- It measures the average attractiveness of attributes across its levels based on how attractive (or unattractive) its full range of levels is vs others
- It measures the gaps between the least and most attractive levels of attributes showing the relative leverage that can be obtained on preference from switching to more attractive levels
- Produces a highly differentiating scale across both options and respondents
- Enables respondents to directly trade-off preferences
This approach is preferred where the goal is to understand which features drive behaviour, rather than the main goal being to simulate preference share for a complete bundle of features. It gives a direct comparison of ALL attribute levels relative to each other – something which isn’t possible in a CBC. It also has greater sensitivity to less important attributes, which can have the potential to be dominated in a regular CBC. We have implemented special simulators for this method.
Although discussed in the choice modelling literature for many years, these models are often underused in practice and we often advocate the use of these as a more economic and effective alternative to a Choice Based Conjoint.