Three essays on the effects of interface and interaction design features on online consumer behaviors

Abstract

As consumers increasingly rely on digital platforms in their consumption, it is crucial to study how various design features of these digital platforms could affect online consumer behaviors. This dissertation presents three essays on the effects of the interface and interaction design features on online consumer behaviors. In essay 1, I study the effect of an interface design factor, namely, display mode (i.e., dark vs. light mode) on consumers’ online fraudulent behaviors. Across three experiments, I found consistent evidence that using a dark mode (vs. light mode) increases consumers’ online fraudulent behaviors by decreasing their moral awareness. The mechanism is tested through both a mediation approach and a moderation approach. I also identify a boundary condition, such that the effect is attenuated when consumers perceive a higher (vs. lower) risk of conducting fraud. In essay 2 and essay 3, I focus on a recent trend of implementing AI agents to interact with consumers and provide services. Specifically, in essay 2, I investigate the dual role of human involvement in AI-enabled customer service. By conducting a field experiment in the customer service system of a commercial bank in China, I found that human involvement is critical in providing both utilitarian and emotional values. Firstly, AI-enabled customer service with partial or full human involvement can provide better problem-solving performance than the service without human involvement. Secondly, compared with AI service with no or partial human involvement, AI service with human involvement in full stage leads customers to perceive the company to be more customer-centric. Lastly, both problem-solving performance and perceived customer orientation are found to significantly increase customers’ intention to use the service in the future. In essay 3, I explore consumers’ bias in evaluating the service provided by human vs. AI agents. Results from a field experiment are consistent with my theorization such that consumers tend to avoid criticizing the human (vs. AI) agent to avoid social discomfort when their problems are not resolved. However, when their problems are resolved, they don't show such leniency in evaluating human (vs. AI) service agents.