BENCHMARKS FOR DISCRETE-CHOICE MODELS

In assessing the performance of a choice model, we have to answer the question, ''Compared with what?'' Analyses of consumer brand choice da ta historically have measured fit by comparing a model's performance w ith that of a naive model that assumes a household's choice probabilit y on each occasion equals the aggregate market share of each brand. Th e authors suggest that this benchmark could form an overly naive point of reference in assessing the fit of a choice model calibrated on sca nner-panel data, or any repeated-measures analysis of choice. They pro pose that fairer benchmarks for discrete choice models in marketing sh ould incorporate heterogeneity in consumer choice probabilities, evide nce for which is by now well documented in the marketing literature. T hey use simulated data to compare the performance of parametric and no nparametric benchmark models, which allow for heterogeneity in consume r choice probabilities, with the performance of the aggregate share-ba sed benchmark model, which assumes consumers are homogeneous in their choice probabilities. They also assess the performance of two previous ly published consumer behavior models against the proposed fairer benc hmark models that allow for heterogeneity in consumer choice probabili ties. They find that one provides a significantly better fit than thei r more conservative benchmark models and the other performs less favor ably.