A Two-Process View of Trust and Distrust Building in Recommendation Agents: A Process-Tracing Study
A Two-Process View of Trust and Distrust Building in Recommendation Agents: A Process-Tracing Study
复制标题
DOI：
10.17705/1jais.00180
复制
发表时间：
2008-01-01
影响因子：
5.8
通讯作者：
Benbasat, Izak
中科院分区：
文献类型：
--
作者：
Komiak, Sherrie Y. X.;Benbasat, Izak
先前的文献关注于信任，而很大程度上忽略了不信任，部分原因是假设在IT中建立信任的信息技术（IT）设计也会阻止不信任的建立。然而，如果IT使用环境中的信任建立过程和不信任建立过程是不同的，那么这个假设可能就不成立了。本文提出了建立信任和不信任的双过程观，即建立信任和不信任的过程是截然不同的。在电子商务中使用推荐代理（RA）的上下文中，信任（不信任）过程被定义为客户对他或她与RA的交互的有利（不利）解释，导致他或她的购物决策可以依赖RA的积极（消极）期望。本研究对过程理论而非方差理论进行了实证检验。方差理论研究依靠逻辑论证来解释和检验变量之间的因果关系。过程理论研究通过揭示和测试构成某些变量影响其他变量的过程的机制来补充方差理论研究。在这项过程追踪研究中，我们收集并分析了49名参与者使用两种RAs的并发口头协议。我们的协议分析结果支持提出的双过程观点。RA使用中的信任建立过程的模式与不信任建立过程的模式有系统的不同，这可能表明RA的一些特征应该被设计为增加信任，而另一些特征应该被设计为减少不信任。研究结果还表明，不信任本身的优点值得研究关注。在一种既涉及建立信任又涉及建立不信任的复杂关系中，理解信任和不信任的过程，而不是只关注信任，可以更准确地描述和改善对这种复杂关系的管理。
Prior literature focuses on trust, while largely ignoring distrust, partly because of the assumption that an Information Technology (IT) design that builds trust in the IT will also prevent distrust-building. However, this assumption may not be true if trust-building processes and distrust-building processes in the context of IT usage are different.This paper proposes a two-process view of trust and distrust building, i.e., that trust-building and distrust-building processes are distinct and separate. In the context of recommendation agent (RA) usage in electronic commerce, a trust (distrust) process is defined as a customer's favorable (unfavorable) interpretation of his or her interactions with an RA, resulting in a positive (negative) expectation that the RA can be relied upon for his or her shopping decisions.This study empirically tests a process theory rather than a variance theory. Variance theory research relies on logical arguments to explain and test the causality relationships among variables. Process theory research complements variance theory research by revealing and testing the mechanisms that constitute the processes by which certain variables influence others. In this process-tracing study, we collected and analyzed the concurrent verbal protocols from 49 participants using two RAs.The results of our protocol analysis support the proposed two-process view. The pattern of trust-building processes in RA usage is systematically different from that of distrust-building processes, which may suggest that some RA features should be designed to increase trust, and others to decrease distrust. The findings also suggest that distrust deserves research attention on its own merit. In a complex relationship involving both trust building and distrust building, understanding both trust and distrust processes, rather than focusing on trust alone, can lead to a more accurate representation and improved management of that complex relationship.