The Effect of Chatbot Communication Style on Academic Stress among University Students: A Randomized Controlled Trial

1. Purpose and Research Problem This research investigates the effectiveness of Generative Artificial Intelligence (genAI) chatbot as a tool for mitigating academic stress among undergraduate students in Indonesia. While chatbots are increasingly used for psychological intervention in mental health context, the "how" of their communication often outweighs the "what" of their content. This study specifically explores how varying levels of Person-Centeredness in chatbot AI communication style, delivered via a WhatsApp-based chatbot named Ciko, and how they impact the emotional recovery of students during digital venting sessions 2. Theoretical Framework The study is grounded in the intersection of two primary psychological frameworks: - Person-Centeredness Theory (Burleson, 2003): This theory posits that effective social support requires the recognition and validation of the other person's emotional state - Transactional Model of Stress and Coping (Lazarus and Folkman, 1984): This frameworks is used to understand how students appraise academic stressors and how external interventions can facilitate more effective emotional-focused coping. 3. Methodology Overview This research employs a randomized experimental design using high-fidelity technical infrastructure to ensure ecological validity. Participants engage in a 7-day "venting" intervention (scheduled for May 2026) where they interact with the Ciko chatbot to discuss their academic challenges - Technical Stack: The experiment is powered by an automated workflow using n8n, WAHA, and Redis for memory management, with the Claude API serving as the core LLM engine. - Experimental Manipulation: One group interacts with a Friendly and Supportive (High Person-Centeredness (HPC)) prompt (validating and empathetic), while the other interacts with Direct and Task-Oriented (Low Person-Centeredness (LPC)) prompt (task-oriented and rigid) - Measurement: Changes in stress levels are tracked using the PSS-10, SISCO Inventory of Academic Stress to compare pre-intervention and post-intervention scores. 4. Expected Outcomes The primary expected outcome is that the Friendly and Supportive (HPC) communication style will result in a significantly greater reduction in academic stress compared to the Direct and Task-Oriented (LPC) style