Incorporating context effects in the multidimensional scaling of 'pick any/N' choice data

This paper presents a multidimensional scaling model that is estimated on pick any/N choice data, and accommodates a broad range of context effects. The methodology estimates a set of parameters capturing the direction and magnitude of the context effects, as well as the locations of brands and consumers' ideal points in a joint multidimensional space. Numerical simulation suggests that our approach can recover the true configuration of brands and ideal points in the presence of context effects with greater accuracy than a 'baseline' model that does not incorporate any context effects. To illustrate the methodology and its implications, we provide an empirical application based on choice of supermarkets, from a consumer study conducted in the Netherlands.