Overview of the MaxDiff Platform
MaxDiff, also known as best-worst scaling (BWS), is a choice-based measurement method. Rather than asking a respondent to report one favorite choice among several alternative profiles, MaxDiff asks a respondent to report both a best and a worst choice. The MaxDiff approach can provide more information about preferences than an approach where a respondent reports only a favorite choice. For background on MaxDiff studies, see Louviere et al. (2015). For background on choice modeling, see Louviere et al. (2015), Train (2009), and Rossi et al. (2005).
MaxDiff analysis uses the framework of random utility theory. A choice is assumed to have an underlying value, or utility, to respondents. The MaxDiff platform estimates these utilities. The MaxDiff platform also estimates the probabilities that a choice is preferred over other choices. This is done using conditional logistic regression. See McFadden (1974).
Note: One-factor MaxDiff studies can be designed using the MaxDiff Design platform. See “MaxDiff Design” in the Design of Experiments Guide.
Segmentation and Bayesian Subject-Level Effects
Market researchers sometimes want to analyze the preference structure for each subject separately in order to see whether there are groups of subjects that behave differently. If there are sufficient data, you can specify “By groups” in the Response Data or you could introduce a Subject identifier as a subject-side model term. This approach, however, is costly if the number of subjects is large. Other segmentation techniques discussed in the literature include Bayesian and mixture methods.
If there are not sufficient data to specify “By groups,” you can segment in JMP by clustering subjects using response data and the Save Gradients by Subject option. The option creates a new data table containing the average Hessian-scaled gradient on each parameter for each subject. For an example, see “Example of Segmentation”. For more information about the gradient values, see “Gradients”.
MaxDiff also provides a Hierarchical Bayesian approach to estimating subject-level effects. This approach can be useful in market segmentation.