Objectives: Statistics anxiety is a prevalent concern among psychology students, potentially impeding their success in this empirical field. Recent literature advocates for ChatGPT as an effective educational virtual assistant. However, empirical investigations into its integration into curricula and its impact on university students' learning outcomes, particularly in statistics, are scarce. This study aims to examine the efficacy and mechanism of ChatGPT, functioning as a virtual assistant, on psychology students' statistics anxiety and academic achievement, while identifying optimal pedagogical strategies for its application. Proposed Methods: Scheduled for the 2024-2025 academic year, this research will occur across two semesters of mandatory statistics courses (PSY3101 and PSY4101). A customized ChatGPT, "PsychStatsGPT", will be introduced for students, incorporating course-specific content and providing course specific assistance. To assess the pedagogical value and efficacy of PsychStatsGPT in reducing students’ statistics anxiety, a randomized controlled experiment, a three-wave longitudinal survey, and focus group interviews will be conducted. The hypothesis posits that PsychStatsGPT, by reducing cognitive load and enhancing support, may diminish statistics anxiety. Impact and Hypothesis: This study promises to deliver comprehensive evidence through mixed methods, enriching theories on ChatGPT's role in educational contexts. It anticipates informing future pedagogical, assessment, teaching methods, curriculum, and policy decisions at Lingnan University and potentially other institutions.