Agent Decision Making: Understanding and Predicting the Preferences of Others

Patricia M. West, University of Texas at Austin
Christina L. Brown, Northwestern University
[ to cite ]:
Patricia M. West and Christina L. Brown (1994) ,"Agent Decision Making: Understanding and Predicting the Preferences of Others", in NA - Advances in Consumer Research Volume 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT : Association for Consumer Research, Pages: 291.

Advances in Consumer Research Volume 21, 1994      Page 291


Patricia M. West, University of Texas at Austin

Christina L. Brown, Northwestern University

Michael Solomon (1986) coined the term "surrogate consumer" when referring to individuals who are employed to make decisions on behalf of their clients. We have broadened the concept to include all individuals who make consumption decisions on another's behalf and refer to these people as "agent decision makers." The notion of a "principal-agent relationship" has been examined in marketing from the economic perspective of agency theory. Rather than the incentive and control issues of traditional agent theory, however, this session focused on the ability of well-intentioned agents to fully understand the principals' preferences and perceptions in order to effectively choose on their behalf.

The first paper, by Patricia M. West, described a laboratory experiment examining agents' ability to learn to predict a principal's tastes. Three factors are identified as impacting agent learning: 1) how the agent responds to feedback from a principal indicating their evaluation of an alternative; 2) the "latitude of acceptance" of the principal (i.e., picky v. easy-to-please); 3) the informational value of feedback from the principal. Information theory suggests that not all feedback is equally informative. A picky person provides informative feedback when he/she reports liking an alternative. An easy-to-please person provides informative feedback when he/she reports disliking an alternative. Subjects are sensitive to differences in the informational value of feedback. Those subjects who focused attention on informative feedback demonstrated higher predictive accuracy. Subjects who viewed mostly preferred alternatives did not learn as well as subjects who viewed either a balance of preferred and nonpreferred, or mostly nonpreferred alternatives.

The second paper, by Christina L. Brown, expanded "agent decision making" to include decisions made by marketers on behalf of consumers. These types of agent decisions are made more complex and problematic by two important features of the marketer's environment: first, marketers are held highly accountable for their decisions and must justify them to others; second, marketers face a more complex set of decision inputs than do consumers. Accountability will change the way the inputs are integrated but may not result in better decisions. This framework was tested on 124 advertising professionals predicting consumer responses to ads in an experiment simulating a typical advertising copy meeting. Results demonstrated that private intuitive responses to ads dominate the decision process regardless of accountability; that accountable decision makers make more systematic, analytical decisions; but that accountability does not always improve decision accuracy. In particular, when market research was not present or was of weak diagnosticity, most decision makers would have made more accurate predictions by relying exclusively on their private intuitive responses to ads.

The third paper, by Deborah Salmond, Jeffery R. Edwards and Robert E. Spekman, examined agent decision making in a business-to-business context. The results from an in-depth field research project were discussed. Key informants on both sides of the buyer-seller relationship participated in interviews and a response questionnaire. Each side reported on its view of the relationship by assessing its own and the other's performance and transaction costs. Successful relationships were ones in which partners had shared expectations and were in consensus with respect to the management of the relationship. Response surface methods were used to assess the effects of shared expectations and consensus on performance and transactions costs.

The concluding discussion by John Deighton addressed the question of whether the concept of agent decision making could be considered "consumer behavior," arguing positively that the prediction of consumers' behavior was an important part of the West and Brown papers. He also suggested that the notion of agent decision making was useful beyond the consumer behavior context, as evidenced by the business-to-business application in the Salmond et al. paper.