This paper addresses the question of how to combine online and offline services in the most complementary way for different product classes. In a series of surveys conducted for Experiment 1 it was determined that consumers ’ preferences for online and offline services differ for different products at different stages of the shopping experience. These differences were accounted for by a model that weights the importance of different attributes for different products and assigns different values to these attributes depending on whether they are better served online or offline. For example, for products like clothing consumers place great value on the ability to touch and inspect the product and thus they prefer offline, bricks-and-mortar services at each stage of the shopping experience. By contrast, for products like computer software consumers place great value on the rapid dissemination of large amounts of information through Internet search, but many are concerned about speedy delivery and nohassle exchange which leads them to make their final purchases offline. Experiment 2 was a controlled test of a particular marketing strategy for capitalizing on the complementarity of online and offline services: alliances between online and offline brands. Confirming the operation of both assimilation and complementarity effects, it was found that the images of both brands could be improved with such alliances. Other marketing strategies were also discussed. 1.

This paper is under journal review. No parts of it may be reproduced or disseminated in any form without permission of the authors.

would like to thank Chris Janiszewski for helpful comments on an earlier version of this paper. 3 Subadditive Bundle Preferences and the Value of Variety Product bundles may consist of multiple units of a single product (“uniform” bundles) or multiple units of more than one product (“mixed ” bundles). In this paper, we examine consumer evaluations of mixed and uniform bundles. Two factors can explain preferences for mixed bundles. First, consumers may perceive each additional unit of a particular product as possessing less marginal value. Second, consumers may derive value from the variety offered by mixed bundles. In study 1, we demonstrate that although consumers do show diminishing marginal utility for uniform bundles of products, they combine products to form mixed bundles in a subadditive fashion. The observedsubadditivityofmixedbundlesisinturnconsistentwithtwoexplanations. Consumers could discount bundles according to the total number of units they contain. Alternatively, consumers could show diminishing marginal utility for product attributes and judge bundles according to the (discounted) value of the attribute inventories in the

acknowledge the helpful input of Alan Cooke. 3 Bundle offers consisting of two or more products often include a price discount. The impact of the price discount on the perceived attractiveness of the bundle has been shown to depend on which product is discounted. It has been argued that discounts are more effective when they are assigned to the product that will receive the most weight in the overall evaluation of the bundle. We propose that the perceived value of the discount may also depend on a referent specific to each product. Six studies are used to provide evidence that (1) price discount framing effects can be explained by reference dependence and (2) that reference dependence and product importance independently contribute to price discount framing effects. 4 Guiltinan (1987, p. 74) broadly defines bundling as “the practice of marketing two or more products and/or services in a single package for a special (i.e., lower) price. ” There are numerous examples of situations where this practice is employed. For example, banks offer lower priced insurance, credit cards, and financial services when these products are purchased as a bundle. Resorts offer lower priced airfare, accommodations, and event tickets when these products are purchased as a bundle. Software companies offer lower prices on suites of software

www.elsevier.com/locate/obhdp Why more can be less: An inference-based explanation for hyper-subadditivity in bundle valuation q