|
Preference heterogeneity aids marketers in crafting a diverse and profitable portfolio of products. Or if just one product in aligning supply with the features target consumers value the most.
Modern choice modeling captures preference heterogeneity via Hierarchical Bayes (HB) parameter estimation. In practice marketers are often disappointed in the lack of preference differences across pre defined groups of consumers. Covariates or Z-variables improve the chances of seeing differences across groups, but alas, do not guarantee any meaningful differences.
We propose a new modeling technique to explore heterogeneity. The standard HB approach models the distribution of heterogeneity in the parameter space. The new model is different; it models heterogeneity in willingness-to-pay (WTP) space. The result is data derived WTP profiles with differing magnitude and grouping of features they value.
Results will be illustrated with simulator output from real consumer choice data. The new technique will be contrasted with the standard approach. We’ll consider the impact on managerial inference, likelihood, and model error.
|
