Read the working paper
INSEAD Working Paper 2013/74/DS
Most of the search literature considers univariate alternatives even though real-world alternatives are typically characterized by multiple attributes. We address this gap by developing a model of search for multiattribute alternatives, focusing on the case of parallel search. The univariate model of parallel search has been applied to vendor selection, new product development, and innovation tournaments; the same model can be applied to a multiattribute setting if the trade-offs to be used at the selection stage are known at the search stage. Yet because parallel search is often used when considerable time elapses between the search and selection stages, we must consider the case where the search decision is made before the trade-offs (to be used at the selection stage) are known. We show that accommodating uncertainty about trade-offs changes the optimal decision at the search stage. In particular, when the distribution of attributes is multivariate normal then more search is needed if uncertainty about trade-offs at the search stage is taken into account. As a result, underappreciation of uncertainty about trade-offs leads to a suboptimal search.