Preference Construction and Persistence in Artificial Marketplaces: The Role of Electronic Recommendation Agents
Authors: Häubl, Gerald and Kyle B. Murray
This paper examines the role of electronic recommendation agents in connection with consumers’ construction of preference for multi-attribute products. A recommendation agent is conceptualized as a software tool that (1) calibrates a model of the subjective preference of a consumer based on his/her input and (2) uses this model to make personalized product recommendations. Based on the notion that preferences tend to be constructive in the sense that they may be affected by characteristics of the task and information environment in which a purchase decision is made, we propose that such digital agents have the potential to influence consumers’ preferences in a systematic fashion. Real-world recommendation systems for online shopping are almost inevitably selective in the sense that they include only a subset of the pertinent product attributes. Our key hypothesis is that, everything else being equal, the inclusion of an attribute in an electronic recommendation agent renders this attribute more prominent in consumers’ purchase decisions. The results of an experiment involving an agent-assisted shopping task provide support for this inclusion effect. We also find that this type of preference-construction effect persists beyond the initial shopping experience and into subsequent preferential choice tasks in which no recommendation agent is available. Finally, we propose three possible explanations of the inclusion effect and discuss each of these in light of our results.