Similarity-based Decision Models∗

This paper introduces similarity-based models of decision making for decision situations in which data on the attributes of options are not available. We propose that an actor, to evaluate a given option, can rely on the utility similar actors derive from this option, or on the utility she derived from similar options. We propose two models to assess the similarity of actors and options. We analyze the proposed models through numerical analyses, and illustrate their workings on a dataset in which reviewers rate restaurants. We nd that the proposed decision making models provide reasonable prediction even if information is not available on the attributes of the options and/or the preferences of the actors. Finally, we discuss the ndings and the implications to other domains of decision making. In reality, all arguments from experience are founded on the similarity which we discover among natural objects, and by which we are induced to expect e ects similar to those which we have found to follow from such objects. (...) From causes which appear similar we expect similar e ects. This is the sum of all our experimental conclusions. David Hume: An Enquiry Concerning Human Understanding, 2004: 21 (1748) ∗This paper bene tted from discussions with Jerker Denrell. All remaining errors are my own. Corresponding address: USI Institute of Management, Via Bu 13, Lugano, Switzerland 6900. Email: [email protected]