Numerous studies of predecisional information purchase indicate that humans are often seriously suboptimal in balancing the costs and benefits of the information they purchase. Underpurchase is reported in some tasks, over-purchase in others, but no convincing account has been offered of the mechanisms producing these patterns of error. The present paper attempts such an account. We first show by computer simulation that simple hill-climbing algorithms reliably reproduce patterns of over-and underpurchase found in earlier studies, and identify the task characteristics that drive this result. We then specify a novel task that should, by this account, yield a new pattern of purchase error and demonstrate that both simulated and real subjects do, in fact, show such error patterns. Implications are drawn both for research strategy in the area and for practical application.
|Original language||English (US)|
|Number of pages||13|
|Journal||Organizational Behavior and Human Decision Processes|
|State||Published - Aug 1988|
Bibliographical noteFunding Information:
Financial support was provided, in part, by O&e of Naval Research Contract 00014-83-K-0742 to the first author. We are grateful to Russ Ferrell, Margaret Neale, Brian Thorn, Gary Wagner, and Bill Waller for helpful comments on earlier drafts. Requests for reprints should be addressed to Terry Connolly, Management and Policy Department, College of Business and Public Administration, University of Arizona, Tucson, AZ 85721.