Enhancing citizen experience with anthropomorphic chatbots in public services: The impact of avatar style and politeness strategy

Abstract

Despite the global trend of integrating chatbots in public services, citizens’ overall acceptance of these automated systems has been low since they may lack human touch and user-friendliness. The enhancement of citizen experience in chatbot-enabled public services has been an under researched topic in public administration. This study conducted a controlled experiment based on 2*2 factorial design to elicit and observe citizens’ interaction with a public service chatbot. The mock chatbot was created with four different combinations of avatar style (human-like vs machine-like) and politeness strategy (positive vs negative). 76 participants were recruited to engage in conversations with the chatbot regarding a typical citizen inquiry, and their evaluations of the chatbot’s social presence, satisfaction levels, and trustworthiness were collected. It was found that both the human-like avatar and positive politeness strategy independently led to significantly higher levels of citizen satisfaction and trust compared to their counterparts. The effects of avatar style and politeness strategy on the citizen experience were both partially mediated by social presence. The findings provided useful insights into the impact of chatbots’ visual and verbal cues, which helped designers and policymakers make informed decisions to create more effective chatbots in public service settings.