AN AGGREGATE CONTEXTUAL CHOICE MODEL FOR ESTIMATING DEMAND FOR NEW PRODUCTS FROM A LABORATORY CHOICE EXPERIMENT

An aggregate contextual choice model (ACCM) is presented and 2 alternative error theories are developed that lead to multinomial maximum likelihood and least-squares estimation procedures. The model is used to estimate demand for 2 new product concepts in a reduced design laboratory choice experiment. The results of the comparison between the ACCM and the Luce constant utility model strongly favor the context-sensitive model over the Luce model. Furthermore, it is shown that the context-sensitive model is easily transformed so that the least-squares method produces results that are very close to maximum likelihood estimates. © 1991.