A nonhierarchical brand switching model for inferring market structure

An aggregate model of market structure based on brand switching data is presented which depicts the market as a set of overlapping clusters of substitutable brands. It generalizes earlier models based on latent class analysis by accounting for heterogeneity among consumers and by allowing for brand switching across clusters-a way of depicting variety seeking. A behavioral rationale for these aggregate models is demonstrated by deriving the aggregate model from an individual level of brand switching. An application to the ground caffeinated coffee market is given.