Multicategory Purchase Incidence with Cross Effects and Co-Incidences

We investigate cross effects of marketing variables and cross category dependences in 25 categories using data on shopping visits of a random sample of 1500 households. We focus on differences between a full model which includes both cross category dependences and cross effects, and a restricted model which ignores cross effects. Estimation gives evidence to several high significant differences of these two models with respect to cross category dependences. A manager using the restricted model would erroneously expect an increase of average basket value due to features (display) in about 28 % (36 %) of the categories.