Extremeness Seeking: When and Why Consumers Prefer the Extremes

John T. Gourville, Dilip Soman
2007 Social Science Research Network  
is Professor of Marketing at the Rotman School of Management, University of Toronto, Ontario, Canada M5S 3E6. We thank Emily Chow and Michael Walker for their research assistance. We also thank Xiuping Li, Amar Cheema, and Mike Norton for comments at various stages of this research. Both authors contributed equally to the research reported in this article. The Compromise Effect Consider a set of alternatives-x, y, and z-that vary on two dimensions, such as price and quality, with x having high
more » ... with x having high price and quality, y having middling price and quality, and z having low price and quality. In the most straightforward demonstrations of the compromise effect, the likelihood of choosing Alternative y increases when the size of the choice set increases from {x, y} to {x, y, z}. In one experiment, for example, Simonson (1989) presented some subjects with two apartments that varied in their general condition and distance from campus, with Apartment x being of high quality and 11 miles from campus and Apartment y being of medium quality and 6 miles from campus. He presented other students with three apartments-Apartment x, Apartment y, and a third apartment, z, that was low in quality and 1 mile from campus. As per the compromise effect, he found that students presented with {x, y} chose Apartment y 50% of the time, while students presented with {x, y, z} chose Apartment y 66% of the time. This and similar findings are interesting at several levels. First, they violate regularity, which argues that the likelihood of choosing an alternative from a given choice set should never increase with the addition of another alternative to that set (Huber, Payne, and Puto 1982; Luce 1959; Tversky 1972) . In other words, the likelihood of choosing Apartment y should not have increased as the choice set expanded from {x, y} to {x, y, z}. Second, this effect appears to be sizeable and robust. Across studies, Simonson (1989) found that an alternative gained an average of 17.5 share points by being a middling alternative within a choice set, while subsequent research has shown similar effects for calculators, cameras, investment portfolios, mouthwashes, personal computers, audio speakers, and rain ponchos (Benartzi and
doi:10.2139/ssrn.987336 fatcat:b2bmqzmf7reojc6vruirccxr6q