Does message interactivity help or hinder the effects of anthropomorphic online chat agents? Compensation vs. expectation effects in organizational websites
Open Access
- Author:
- Go, Eun
- Graduate Program:
- Mass Communications
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 28, 2015
- Committee Members:
- S. Shyam Sundar, Dissertation Advisor/Co-Advisor
S. Shyam Sundar, Committee Chair/Co-Chair
Denise Sevick Bortree, Committee Member
Fuyuan Shen, Committee Member
Edgar Paul Yoder, Committee Member - Keywords:
- Online chatting agent
compensation effects
expectancy violation effect
visual cue
identity cue
message interactivity - Abstract:
- The popularity of providing customer service via online chat assistants is on the rise. Chat agents provide numerous benefits, including low cost and ease of use. However, using online chat agent, especially, a chat-bot poses challenges as well, in that their conversation styles are somewhat limited and impersonal. Given that chat-bots are designed to substitute for actual human agents, this study suggests that the message interactivity that is characteristic of human-to-human conversation may be one means for improving chat-bots’ communication. In particular, this study aims to investigate how the ability of chat-bots to exchange messages in a contingent manner may compensate for their lack of humanness. In order to investigate this effect, this study conceptualizes the ability to exchange messages in a contingent manner as message interactivity and humanness in terms of the chat agents’ visual representation (as anthropomorphic or not) and role identity (as chat-bot or human). A human-like conversation style of a chat-bot is expected to compensate for the chat-bot’s lack of its humanness; however, it is also possible that if its conversation style is too contingent in style, and thus too human-like, an expectancy violation may occur. Thus, this study pays special attention to these two possibilities (i.e., the compensation effect and the expectancy violation effect) to explain the effects of message interactivity using the Interactivity Effects Model (Sundar, 2007) and the theory of interactive media effects (TIME) proposed by Sundar et al. (2015b). These possibilities were investigated with a 2 (anthropomorphic visual cue: high vs low anthropomorphism) × 2 (identity cue: chat-bot vs. sales associate) × 2 (message interactivity: high vs. low message interactivity) between-subjects experiment. Participants were asked to interact with an online chat agent by using “live chat” function on an e-commerce website. By examining the interactions of visual cue, identity cue and message interactivity, this study revealed distinctive mechanisms through which message interactivity affects users’ psychological, attitudinal, behavioral and relational responses. Specifically, findings show that message interactivity works to compensate for the impersonal nature associated with low anthropomorphic visual cue and chat-bot identity cue. Moreover, the identity cue turned out to be a key factor in eliciting certain expectations regarding the agent’s performance in conversation. Therefore, this study suggests using chat agents on organizational websites in order to help mitigate users’ negative evaluations or experiences due to high expectations given that users tend to have different expectations of agents’ performances, depending on whether the agents are human or chat-bots. Therefore, agents’ actual communication styles should align with users’ expectations. More theoretical as well as practical implications of these findings are discussed.